final.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# <a id='toc1_'></a>[Synergy Transfer](#toc0_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Author: Blake McBride<br>\n",
    "Email: blakepm2@illinois.edu<br>\n",
    "Course: CS 590 Deep Learning for Healthcare<br>\n",
    "Professor: Jimeng Sun<br>\n",
    "\n",
    "This notebook presents the reproduction results for the paper **\"Anticancer drug synergy prediction in understudied tissues using transfer learning\"** ([Kim et al. 2020](https://doi.org/10.1101/2020.02.05.932657)). All major code components have been reused and adapted from the work of the original authors, with modifications made for reproduction purposes.\n",
    "\n",
    "The original publication can be accessed [here](https://academic.oup.com/jamia/article-abstract/28/1/42/5920819?redirectedFrom=fulltext&login=false).\n",
    "\n",
    "The original publication GitHub repoitory can be accessed [here](https://github.com/yejinjkim/synergy-transfer).\n",
    "\n",
    "The public GitHub repository for this project can be accessed [here](https://github.com/blakepm2/synergy_transfer)\n",
    "\n",
    "The video presentation for this project can be found [here](https://www.youtube.com/watch?v=viN0ZtUBl_g)."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<a href=\"https://www.youtube.com/watch?v=viN0ZtUBl_g\">\n",
    "<div align=\"center\">\n",
    "    <img src=\"public/assets/thumbnail.png\" width=\"700\">\n",
    "</div>\n",
    "</a>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## <a id='toc1_1_'></a>[Table of Contents](#toc0_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    " - [Synergy Transfer](#toc1_)    \n",
    "  - [Table of Contents](#toc1_1_)    \n",
    "  - [Introduction](#toc1_2_)    \n",
    "  - [Scope of Reproducability](#toc1_3_)    \n",
    "    - [A Note on Previous Work: MoleculeSTM](#toc1_3_1_)    \n",
    "      - [Environment Complexity](#toc1_3_1_1_)    \n",
    "      - [Computational Limitations](#toc1_3_1_2_)    \n",
    "    - [Switching to Synergy Transfer](#toc1_3_2_)    \n",
    "      - [Absences and Misalignments in Data](#toc1_3_2_1_)    \n",
    "  - [Methodology](#toc1_4_)    \n",
    "    - [Environment](#toc1_4_1_)    \n",
    "      - [Conda](#toc1_4_1_1_)    \n",
    "      - [Python Dependencies](#toc1_4_1_2_)    \n",
    "    - [Data](#toc1_4_2_)    \n",
    "      - [Sources](#toc1_4_2_1_)    \n",
    "      - [Configuration](#toc1_4_2_2_)    \n",
    "      - [Download](#toc1_4_2_3_)    \n",
    "      - [Summary](#toc1_4_2_4_)    \n",
    "        - [Data Exploration](#toc1_4_2_4_1_)    \n",
    "        - [Transforming Items to ID](#toc1_4_2_4_2_)    \n",
    "      - [Drugs](#toc1_4_2_5_)    \n",
    "        - [Preprocessing](#toc1_4_2_5_1_)    \n",
    "        - [Aligning Drug's Target Gene(s)](#toc1_4_2_5_2_)    \n",
    "      - [Cells](#toc1_4_2_6_)    \n",
    "        - [Preprocessing](#toc1_4_2_6_1_)    \n",
    "        - [Aligning Cell Lines with Gene Expression(s)](#toc1_4_2_6_2_)    \n",
    "        - [Gene Integration](#toc1_4_2_6_3_)    \n",
    "        - [Data Exploration](#toc1_4_2_6_4_)    \n",
    "    - [Model](#toc1_4_3_)    \n",
    "      - [Architecture](#toc1_4_3_1_)    \n",
    "        - [Drug Encoder](#toc1_4_3_1_1_)    \n",
    "        - [Cell Line Encoder](#toc1_4_3_1_2_)    \n",
    "        - [Merging Layers](#toc1_4_3_1_3_)    \n",
    "      - [Configuration](#toc1_4_3_2_)    \n",
    "      - [Gene Compression](#toc1_4_3_3_)    \n",
    "        - [Dataset and Dataloader](#toc1_4_3_3_1_)    \n",
    "        - [Compress gene features](#toc1_4_3_3_2_)    \n",
    "      - [Synergy prediction](#toc1_4_3_4_)    \n",
    "        - [Train/test split in cross or external validation](#toc1_4_3_4_1_)    \n",
    "        - [Dataset and Dataloader](#toc1_4_3_4_2_)    \n",
    "        - [Load dataset](#toc1_4_3_4_3_)    \n",
    "      - [Prediction model](#toc1_4_3_5_)    \n",
    "      - [Training](#toc1_4_3_6_)    \n",
    "        - [Hyperparameters](#toc1_4_3_6_1_)    \n",
    "        - [Computational Requirements](#toc1_4_3_6_2_)    \n",
    "      - [Evaluation](#toc1_4_3_7_)    \n",
    "        - [Evaluation Metrics](#toc1_4_3_7_1_)    \n",
    "      - [Train Specific Model](#toc1_4_3_8_)    \n",
    "        - [Freeze layers](#toc1_4_3_8_1_)    \n",
    "      - [Evaluate Specific Model](#toc1_4_3_9_)    \n",
    "        - [Select the top ranked drug combinations](#toc1_4_3_9_1_)    \n",
    "  - [Results](#toc1_5_)    \n",
    "    - [Drug response prediction accuracy for data-rich tissues](#toc1_5_1_)    \n",
    "    - [Drug response prediction accuracy for bone and prostate cancer](#toc1_5_2_)    \n",
    "    - [Ranking Predicted Synergenistic Drug Combinations](#toc1_5_3_)    \n",
    "      - [Prostate Top Predictions](#toc1_5_3_1_)    \n",
    "      - [Bone Top Predictions](#toc1_5_3_2_)    \n",
    "  - [Discussion](#toc1_6_)    \n",
    "    - [Issues Encountered](#toc1_6_1_)    \n",
    "    - [Reproducability and Future Considerations](#toc1_6_2_)    \n",
    "  - [References](#toc1_7_)    \n",
    "\n",
    "<!-- vscode-jupyter-toc-config\n",
    "\tnumbering=false\n",
    "\tanchor=true\n",
    "\tflat=false\n",
    "\tminLevel=1\n",
    "\tmaxLevel=6\n",
    "\t/vscode-jupyter-toc-config -->\n",
    "<!-- THIS CELL WILL BE REPLACED ON TOC UPDATE. DO NOT WRITE YOUR TEXT IN THIS CELL -->"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## <a id='toc1_2_'></a>[Introduction](#toc0_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In the pursuit of enhancing the efficacy of cancer treatments, the exploration of synergistic drug combinations has emerged as a promising strategy. Indeed, Kim et al. demonstrate in their paper how high-throughput combinatorial drug screening offers a pathway to discover these potent combinations, leveraging the complex interplay between multiple drugs to achieve higher efficacy without compromising safety ([Kim et al. 2020](https://doi.org/10.1101/2020.02.05.932657)). Whereas traditional drug discovery processes are not only costly and time-consuming but often result in therapies that target single molecules or pathways, potentially missing opportunities for enhanced therapeutic effects through drug combinations.\n",
    "\n",
    "A significant challenge in this domain is the uneven distribution of data across various cancer types. While some cancers have been extensively studied, providing a rich dataset of drug responses, others, such as bone and prostate cancers, remain understudied due to various obstacles including the difficulty in obtaining viable cell lines and the complexity of their in-vitro models ([Kim et al. 2020](https://doi.org/10.1101/2020.02.05.932657)). This disparity poses a problem for developing effective treatments for these less-studied tissues, as the lack of data hampers the predictive power of drug response models. Addressing this issue, Kim et al. present a novel approach to predicting effective drug combinations in tissues with scant data. By employing **transfer learning**, the study leverages the abundance of data from well-studied tissues to enhance predictions in data-poor tissues, thus overcoming a significant hurdle in cancer treatment research. The methodology centers around the development of a deep neural network model that integrates a comprehensive set of genetic, molecular, and phenotypic features collected from various databases. This model not only predicts drug synergy but also adapts the knowledge learned from data-rich contexts to those that are data-deficient.\n",
    "\n",
    "The core innovation of this study lies in its use of transfer learning, a technique where a model developed for one task is reused as the starting point for a model on a second task. Here, it involves transferring the insights gained from drug-cell interactions in common cancers to improve synergy predictions in less common ones. This approach is particularly advantageous in dealing with the sparse data scenarios typical of understudied tissues. By doing so, the model provides a means to prioritize future in-vitro experiments, potentially accelerating the discovery of viable cancer treatments for these challenging cases.\n",
    "\n",
    "The work of Kim et al. significantly contributes to the field of computational drug discovery by demonstrating how advanced machine learning techniques can bridge the gap between data-rich and data-poor scenarios. This not only enhances the predictive accuracy of drug synergy models but also broadens the scope of their applicability to a wider range of tissues, ultimately facilitating more targeted and effective cancer treatments."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## <a id='toc1_3_'></a>[Scope of Reproducability](#toc0_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In terms of the reproducability of this paper, I encountered several issues while attempting to reproduce the authors' results ([Kim et al. 2020](https://doi.org/10.1101/2020.02.05.932657)). I should make note here that this has been a consistent issue I've faced throughout the course of this project, as this is the 2nd project that I've attempted. To demonstrate the sincere efforts that have been put into this project without being able to showcase any final results, I will detail my process here."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### <a id='toc1_3_1_'></a>[A Note on Previous Work: MoleculeSTM](#toc0_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Originally, I chose to reproduce the results of [Chao et al. (2023)](https://doi.org/10.1038/s42256-023-00759-6) for my assignment. In their work, Chao et al. premiered MoleculeSTM, a multi-modal molecule structure-text model for text-based retrieval and editing. The goal of their work was to showcase a new machine learning approach for drug discovery that, in separation from other work in this domain, would jointly learn molecules chemical structures as well as their textual descriptions through a contrastive learning approach ([Chao et al. 2023](https://doi.org/10.1038/s42256-023-00759-6)). I was unable to reproduce the results of this paper for two primary reasons: the complexity of environmental dependencies, and the computational feasibility."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### <a id='toc1_3_1_1_'></a>[Environment Complexity](#toc0_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Regarding the environment setup process, this proved to be very taxing and took nearly a week to successfully complete. This seems to stem from a dependency conflict between some of the packages used, namely between PyTorch, CUDA, and Apex. I tried to configure the environment on three different operating systems; starting with Windows 11, then MacOS Sonoma, and finally Ubuntu 22.04. I reached out to ShangChao Liu for support and his advice was to set the environment up on Linux, so I partitioned my hard drive and installed Ubuntu, at which point I was able to get through the environment setup successfully. Additionally, after noticing that others were experiencing similar difficulties with the environment setup, I recieved permission from Dr. Liu to make a pull request to the respository with my fix, which explicitly set a compatible version of CUDA binaries to be installed with the PyTorch installation. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### <a id='toc1_3_1_2_'></a>[Computational Limitations](#toc0_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The second issue that made this paper difficult to reproduce within the scope of this project was the computational feasibility. Using an NVIDIA GTX 4090 GPU, I successfully started the pretraining process on Friday, April 12th, and left it running for 2 weeks. I started having concerns after seeing minimal progress the next morning, so I started researching alternative papers from a collection of backups that I kept. After letting the pretraining process run, I was disheartened to see that only 30/100 epochs had completed. I again reached out to Dr. Liu for guidance and he informed me that he and his colleagues originally trained MoleculeSTM on a computing cluster of multiple machines each outfitted with several CPUs and at least 4 NVIDIA 2080 GPUs. This was not a realistic or feasible paper for me to reproduce in time for submission without spending a significant amount of money on accessing higher-powered computing space. Unfortunately, without realizing how long the pretraining process was estimated to take, I neglected to add code to save checkpoints as it went along, which means that after terminating the process I had nothing to show for this. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### <a id='toc1_3_2_'></a>[Switching to Synergy Transfer](#toc0_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Deciding to switch projects to one that would be more feasible required me to examine my backup papers for one that had a lower computational complexity and had readily accessible data, as I did not anticipate a need to request access to any of the MIMIC data given that MoleculeSTM had its data stored in HuggingFace repositories. To this end, I landed on Synergy Transfer by [Kim et al. (2020)](https://doi.org/10.1101/2020.02.05.932657), which seemed promising due to its comparatively lower system requirements and open source data usage. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### <a id='toc1_3_2_1_'></a>[Absences and Misalignments in Data](#toc0_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "While my shift from MoleculeSTM to Synergy Transfer was done primarily for its use of publicly available data, I quickly found that this was a significant challenge due to the unavailability of some of the original data used and the lack of documentation for acquiring it or replicating their findings. Without documentation of any kind, I followed the only indications I had from the original paper as well as the GitHub repo, which stated that data had been accessed from DrugComb and DrugBank. It took time to understand how to acquire the data using the API from DrugComb, but once I did manage to get the data I noticed that there had been significant changes to the data architecture since this paper was firts published. Specifically, key dimensions reflecting the tissue and disease names, as well as the synergy scores for various blocks were absent from the new version of data available on DrugComb. I refactored the code as best as I could to align with the new dimensionality, but there were still issues with the absence of tissue and disease names that caused downsteam errors in the Synergy Prediction task, as will be seen in future sections. \n",
    "\n",
    "Where DrugBank was concerned, I had no realistic pathway to help me acquire it, so I similarly decided to reach out to two of the original authors of the paper, Kim Yejin and Jing Tang, for guidance. Kim Yejin shared with me a .zip file containing some of the data used for downstream tasks in the preprocessing notebook, but it was not inclusive of all of the original data they used. When I requested all of the original data used, she did not respond. Jing Tang shared the original summary .csv file that I was looking for, but the dimensionality of this data did not match the newer version or the version that they used in the original paper, so I was still blocked from fully realizing their results."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## <a id='toc1_4_'></a>[Methodology](#toc0_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This section will cover the methodology used for reproducing the results of [Kim et al. (2020)](https://doi.org/10.1101/2020.02.05.932657), including a full setup guide for the environment, an overview of available data, the model architecture, training process, and evaluation. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### <a id='toc1_4_1_'></a>[Environment](#toc0_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Before beginning the environment setup, I should note that I used a machine with Ubuntu v22.04 installed and would recommend doing the same to avoid some complexity with package installation. Assuming you're working on some distribution of Linux or want to attempt the setup with a different OS, the first step is to install Conda."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### <a id='toc1_4_1_1_'></a>[Conda](#toc0_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "First, install Conda as shown below, you can refer to more comprehensive instructions on [Conda's website](https://docs.conda.io/projects/conda/en/latest/user-guide/install/linux.html) if you're using a different OS.\n",
    "\n",
    "```\n",
    "mkdir -p ~/miniconda3\n",
    "wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh -O ~/miniconda3/miniconda.sh\n",
    "bash ~/miniconda3/miniconda.sh -b -u -p ~/miniconda3\n",
    "rm -rf ~/miniconda3/miniconda.sh\n",
    "```\n",
    "\n",
    "Then initialize bash and zsh shells\n",
    "\n",
    "```\n",
    "~/miniconda3/bin/conda init bash\n",
    "~/miniconda3/bin/conda init zsh\n",
    "```\n",
    "\n",
    "Create a new conda environment using **Python 3.8** as the base. To make things easier, you can use the following command to create a new conda environment using the [environment.yaml](environment.yaml) file, as shown below:\n",
    "\n",
    "```\n",
    "conda env create -f environment.yaml\n",
    "```\n",
    "\n",
    "Or, if you wish to do it yourself, create your new conda environment as shown below:\n",
    "\n",
    "```\n",
    "conda create -n synergy python=3.8\n",
    "```\n",
    "\n",
    "and activate it\n",
    "\n",
    "```\n",
    "conda activate synergy\n",
    "```"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### <a id='toc1_4_1_2_'></a>[Python Dependencies](#toc0_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "With a conda environment successfully configured, you now must install all of the required Python dependencies.\n",
    "\n",
    "```\n",
    "pip install torch torchvision\n",
    "pip install cudatoolkit==12.1\n",
    "pip install pandas\n",
    "pip install scikit-learn\n",
    "pip install seaborn\n",
    "pip install numpy\n",
    "pip install matplotlib\n",
    "pip install rdkit\n",
    "```\n",
    "\n",
    "At this point you should be all set up to start running the code in this notebook. Ensure you select the conda environment you created as the Kernel."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### <a id='toc1_4_2_'></a>[Data](#toc0_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Once the environment has been successfully configured, the next step is accessing and preprocessing the data. The section below provides download instructions for all of the data used for this project (both raw and processed). You can run the code below to perform all preprocessing steps manually, or you can simply use the provided processed data used for model training and evaluation."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### <a id='toc1_4_2_1_'></a>[Sources](#toc0_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The data used for this analysis comes from 4 different sources: \n",
    "\n",
    "- **DrugComb**\n",
    "\n",
    "    - For drug sensitivity and synergy data [(Zagidullin et al. 2019)](https://doi.org/10.1093/nar/gkz337)\n",
    "\n",
    "- **DrugBank**\n",
    "\n",
    "    - For drug genomic features [(Wishart et al. 2007)](https://doi.org/10.1093/nar/gkm958)\n",
    "\n",
    "- **Therapeutic Targets Database (TTD)**\n",
    "\n",
    "    - For drug genomic features [(Chen et al. 2002)](https://doi.org/10.1093/nar/30.1.412)\n",
    "\n",
    "- **NIH-LINC**\n",
    "\n",
    "    - For drug genomic features [(Harvard Medical School 2017)](http://lincsportal.ccs.miami.edu/datasets/#/view/LDS-1415)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### <a id='toc1_4_2_2_'></a>[Configuration](#toc0_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Import required modules"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import pickle\n",
    "import matplotlib.pyplot as plt\n",
    "from mpl_toolkits.mplot3d import Axes3D\n",
    "from matplotlib import cm\n",
    "import rdkit\n",
    "from rdkit import Chem\n",
    "from rdkit.Chem import MACCSkeys\n",
    "import re\n",
    "import seaborn as sns\n",
    "from matplotlib.colors import LogNorm\n",
    "import math\n",
    "import gdown\n",
    "import os\n",
    "\n",
    "from utilities import Mapping"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Set your data path as the directory in which you saved the downloaded data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "DATA_PATH = 'synergy_transfer_data/'"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### <a id='toc1_4_2_3_'></a>[Download](#toc0_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Run the code below to download all of the data to your data path"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Retrieving folder contents\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Retrieving folder 1tJjbItceWewAFKT6GbkiEc0uetG9v66Y cell\n",
      "Retrieving folder 1McFcm50uxfOhSLdV32mpwi1y64WEbyLl processed\n",
      "Processing file 1wn6X6CKzBorf74qYfj299kGjcQh6qcC- cell_features.p\n",
      "Retrieving folder 1DFpAvjQkXsSeR-AQwQZJ_ZQWXMh7Udfn raw\n",
      "Processing file 1vHCitsb1wwjZpb5d5Oph6zefuOImINP1 cell_line_meta.csv\n",
      "Processing file 1WEhqtuSDjZgPAqFBdcYY8d1-pUqY_96s cell_lines.csv\n",
      "Retrieving folder 1NXmg7tbKZP2anOXrJ4dhDbIToFiHUDrx drug\n",
      "Retrieving folder 1uZutTWVUKDwYlkBQTDvCTwRnIKOQeO4s processed\n",
      "Processing file 1OQYj57LwSqJgRXNLHwCeLIo_4E_6g9kW drug_clean_fp.p\n",
      "Processing file 1TZ1v9l4GIQO8fchQPJFEpZHs-Q0LvaP9 drug_clean.csv\n",
      "Processing file 1ZO-Tzwxz1lUYU8RuhWu3G5ZHYwI32cQm drug_features.p\n",
      "Retrieving folder 1DX9dWJPOq3IzyELCH719FXGN92Rjddah raw\n",
      "Processing file 18mZ8ywSYDuUNgeG03YSpznvwdeA68xcy drug_DrugBank_target.csv\n",
      "Processing file 1rIs8Mlm9BJQCyWI1l0QT9h9bn4AdZvsc drug_lincs_target.csv\n",
      "Processing file 1ItbIrk0bpqGEIdfv1S_DC8wUg4I6tasX drug_TTD_target.csv\n",
      "Processing file 1BwIuZb2JhBs07lPOEtg5fRfbjldXBRGB drugs.csv\n",
      "Retrieving folder 14SpivIgJRrLnlHn1QYIgcGd_flPaB1J9 gene\n",
      "Retrieving folder 1OZk6L6URo70u5XOfjAjteSFTkXjzAu4p raw\n",
      "Processing file 1P2EJU5QoLz5X0ChiYwsIeDPLUYCaCWpn fpkm_broad.csv\n",
      "Processing file 1cD6s-mEoaSLzF6kSkIkYhlpoAKY430sC fpkm_sangar.csv\n",
      "Processing file 1euEjVnSsYl_FC32NxwHaxzuuvh0pL7BM gene_identifiers_latest.csv\n",
      "Retrieving folder 1trZ5MTycQYS30WH2krNktqeAfPSLDKVf models\n",
      "Processing file 1d2MOuFBIMffUHd4rOAwhvuVlC_ncHMYA cellGeneCompressed.p\n",
      "Processing file 1YGf5KJr4DmAIbWpzdTW39XMeObdwdtLd cellGeneCompressor.p\n",
      "Processing file 1KpkX68hk7wOLZ2pdYy76eE9dvV1TSAZ0 drugGeneCompressed.p\n",
      "Processing file 1JbkxIOVFvd1LJOQr8bZ6yFzobguKteM6 drugGeneCompressor.p\n",
      "Processing file 1bHutzJt4Rh9SiJTkwYA5SYmbotn34-P1 id_feat_gene.p\n",
      "Retrieving folder 18nsrwRTRk6OAgwXTZrfydsoAlTEsif8U summary\n",
      "Retrieving folder 1Q6dGgNAgQsSpUh3JI6tYnj0-x1MnFs7x processed\n",
      "Processing file 1JgoRl74VF7-53lSAMoHgKMc8wozBZAtj codes.p\n",
      "Processing file 1BC18VDC2Zt9meWrqpw37SlXJNWceAFQN summary_mean.p\n",
      "Retrieving folder 1lUoGN9JQs0e1_y-JbJkPqxTwHr9SJvIs raw\n",
      "Processing file 158U6k-BN-ANkufnFejjDUBNQO9m8oymR summary_table.csv\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Retrieving folder contents completed\n",
      "Building directory structure\n",
      "Building directory structure completed\n",
      "Downloading...\n",
      "From: https://drive.google.com/uc?id=1wn6X6CKzBorf74qYfj299kGjcQh6qcC-\n",
      "To: /home/blake/synergy_transfer/synergy_transfer_data/cell/processed/cell_features.p\n",
      "100%|██████████| 43.7M/43.7M [00:01<00:00, 27.8MB/s]\n",
      "Downloading...\n",
      "From: https://drive.google.com/uc?id=1vHCitsb1wwjZpb5d5Oph6zefuOImINP1\n",
      "To: /home/blake/synergy_transfer/synergy_transfer_data/cell/raw/cell_line_meta.csv\n",
      "100%|██████████| 256k/256k [00:00<00:00, 6.59MB/s]\n",
      "Downloading...\n",
      "From: https://drive.google.com/uc?id=1WEhqtuSDjZgPAqFBdcYY8d1-pUqY_96s\n",
      "To: /home/blake/synergy_transfer/synergy_transfer_data/cell/raw/cell_lines.csv\n",
      "100%|██████████| 259k/259k [00:00<00:00, 7.73MB/s]\n",
      "Downloading...\n",
      "From: https://drive.google.com/uc?id=1OQYj57LwSqJgRXNLHwCeLIo_4E_6g9kW\n",
      "To: /home/blake/synergy_transfer/synergy_transfer_data/drug/processed/drug_clean_fp.p\n",
      "100%|██████████| 2.29M/2.29M [00:00<00:00, 16.3MB/s]\n",
      "Downloading...\n",
      "From: https://drive.google.com/uc?id=1TZ1v9l4GIQO8fchQPJFEpZHs-Q0LvaP9\n",
      "To: /home/blake/synergy_transfer/synergy_transfer_data/drug/processed/drug_clean.csv\n",
      "100%|██████████| 762k/762k [00:00<00:00, 18.6MB/s]\n",
      "Downloading...\n",
      "From: https://drive.google.com/uc?id=1ZO-Tzwxz1lUYU8RuhWu3G5ZHYwI32cQm\n",
      "To: /home/blake/synergy_transfer/synergy_transfer_data/drug/processed/drug_features.p\n",
      "100%|██████████| 1.98M/1.98M [00:00<00:00, 25.4MB/s]\n",
      "Downloading...\n",
      "From: https://drive.google.com/uc?id=18mZ8ywSYDuUNgeG03YSpznvwdeA68xcy\n",
      "To: /home/blake/synergy_transfer/synergy_transfer_data/drug/raw/drug_DrugBank_target.csv\n",
      "100%|██████████| 4.22M/4.22M [00:00<00:00, 27.3MB/s]\n",
      "Downloading...\n",
      "From: https://drive.google.com/uc?id=1rIs8Mlm9BJQCyWI1l0QT9h9bn4AdZvsc\n",
      "To: /home/blake/synergy_transfer/synergy_transfer_data/drug/raw/drug_lincs_target.csv\n",
      "100%|██████████| 787k/787k [00:00<00:00, 16.4MB/s]\n",
      "Downloading...\n",
      "From: https://drive.google.com/uc?id=1ItbIrk0bpqGEIdfv1S_DC8wUg4I6tasX\n",
      "To: /home/blake/synergy_transfer/synergy_transfer_data/drug/raw/drug_TTD_target.csv\n",
      "100%|██████████| 296k/296k [00:00<00:00, 12.2MB/s]\n",
      "Downloading...\n",
      "From: https://drive.google.com/uc?id=1BwIuZb2JhBs07lPOEtg5fRfbjldXBRGB\n",
      "To: /home/blake/synergy_transfer/synergy_transfer_data/drug/raw/drugs.csv\n",
      "100%|██████████| 734k/734k [00:00<00:00, 17.5MB/s]\n",
      "Downloading...\n",
      "From: https://drive.google.com/uc?id=1P2EJU5QoLz5X0ChiYwsIeDPLUYCaCWpn\n",
      "To: /home/blake/synergy_transfer/synergy_transfer_data/gene/raw/fpkm_broad.csv\n",
      "100%|██████████| 56.0M/56.0M [00:01<00:00, 30.3MB/s]\n",
      "Downloading...\n",
      "From: https://drive.google.com/uc?id=1cD6s-mEoaSLzF6kSkIkYhlpoAKY430sC\n",
      "To: /home/blake/synergy_transfer/synergy_transfer_data/gene/raw/fpkm_sangar.csv\n",
      "100%|██████████| 33.7M/33.7M [00:01<00:00, 23.8MB/s]\n",
      "Downloading...\n",
      "From: https://drive.google.com/uc?id=1euEjVnSsYl_FC32NxwHaxzuuvh0pL7BM\n",
      "To: /home/blake/synergy_transfer/synergy_transfer_data/gene/raw/gene_identifiers_latest.csv\n",
      "100%|██████████| 2.91M/2.91M [00:00<00:00, 33.1MB/s]\n",
      "Downloading...\n",
      "From: https://drive.google.com/uc?id=1d2MOuFBIMffUHd4rOAwhvuVlC_ncHMYA\n",
      "To: /home/blake/synergy_transfer/synergy_transfer_data/models/cellGeneCompressed.p\n",
      "100%|██████████| 57.5k/57.5k [00:00<00:00, 4.82MB/s]\n",
      "Downloading...\n",
      "From: https://drive.google.com/uc?id=1YGf5KJr4DmAIbWpzdTW39XMeObdwdtLd\n",
      "To: /home/blake/synergy_transfer/synergy_transfer_data/models/cellGeneCompressor.p\n",
      "100%|██████████| 25.0M/25.0M [00:00<00:00, 30.3MB/s]\n",
      "Downloading...\n",
      "From: https://drive.google.com/uc?id=1KpkX68hk7wOLZ2pdYy76eE9dvV1TSAZ0\n",
      "To: /home/blake/synergy_transfer/synergy_transfer_data/models/drugGeneCompressed.p\n",
      "100%|██████████| 1.06M/1.06M [00:00<00:00, 18.3MB/s]\n",
      "Downloading...\n",
      "From: https://drive.google.com/uc?id=1JbkxIOVFvd1LJOQr8bZ6yFzobguKteM6\n",
      "To: /home/blake/synergy_transfer/synergy_transfer_data/models/drugGeneCompressor.p\n",
      "100%|██████████| 12.6M/12.6M [00:00<00:00, 24.9MB/s]\n",
      "Downloading...\n",
      "From: https://drive.google.com/uc?id=1bHutzJt4Rh9SiJTkwYA5SYmbotn34-P1\n",
      "To: /home/blake/synergy_transfer/synergy_transfer_data/models/id_feat_gene.p\n",
      "100%|██████████| 2.22M/2.22M [00:00<00:00, 30.6MB/s]\n",
      "Downloading...\n",
      "From: https://drive.google.com/uc?id=1JgoRl74VF7-53lSAMoHgKMc8wozBZAtj\n",
      "To: /home/blake/synergy_transfer/synergy_transfer_data/summary/processed/codes.p\n",
      "100%|██████████| 512k/512k [00:00<00:00, 10.9MB/s]\n",
      "Downloading...\n",
      "From: https://drive.google.com/uc?id=1BC18VDC2Zt9meWrqpw37SlXJNWceAFQN\n",
      "To: /home/blake/synergy_transfer/synergy_transfer_data/summary/processed/summary_mean.p\n",
      "100%|██████████| 21.9M/21.9M [00:00<00:00, 29.2MB/s]\n",
      "Downloading...\n",
      "From (original): https://drive.google.com/uc?id=158U6k-BN-ANkufnFejjDUBNQO9m8oymR\n",
      "From (redirected): https://drive.google.com/uc?id=158U6k-BN-ANkufnFejjDUBNQO9m8oymR&confirm=t&uuid=d1c89008-38fb-495d-abcd-3cfa0390a087\n",
      "To: /home/blake/synergy_transfer/synergy_transfer_data/summary/raw/summary_table.csv\n",
      "100%|██████████| 166M/166M [00:05<00:00, 29.2MB/s] \n",
      "Download completed\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "['/home/blake/synergy_transfer/synergy_transfer_data/cell/processed/cell_features.p',\n",
       " '/home/blake/synergy_transfer/synergy_transfer_data/cell/raw/cell_line_meta.csv',\n",
       " '/home/blake/synergy_transfer/synergy_transfer_data/cell/raw/cell_lines.csv',\n",
       " '/home/blake/synergy_transfer/synergy_transfer_data/drug/processed/drug_clean_fp.p',\n",
       " '/home/blake/synergy_transfer/synergy_transfer_data/drug/processed/drug_clean.csv',\n",
       " '/home/blake/synergy_transfer/synergy_transfer_data/drug/processed/drug_features.p',\n",
       " '/home/blake/synergy_transfer/synergy_transfer_data/drug/raw/drug_DrugBank_target.csv',\n",
       " '/home/blake/synergy_transfer/synergy_transfer_data/drug/raw/drug_lincs_target.csv',\n",
       " '/home/blake/synergy_transfer/synergy_transfer_data/drug/raw/drug_TTD_target.csv',\n",
       " '/home/blake/synergy_transfer/synergy_transfer_data/drug/raw/drugs.csv',\n",
       " '/home/blake/synergy_transfer/synergy_transfer_data/gene/raw/fpkm_broad.csv',\n",
       " '/home/blake/synergy_transfer/synergy_transfer_data/gene/raw/fpkm_sangar.csv',\n",
       " '/home/blake/synergy_transfer/synergy_transfer_data/gene/raw/gene_identifiers_latest.csv',\n",
       " '/home/blake/synergy_transfer/synergy_transfer_data/models/cellGeneCompressed.p',\n",
       " '/home/blake/synergy_transfer/synergy_transfer_data/models/cellGeneCompressor.p',\n",
       " '/home/blake/synergy_transfer/synergy_transfer_data/models/drugGeneCompressed.p',\n",
       " '/home/blake/synergy_transfer/synergy_transfer_data/models/drugGeneCompressor.p',\n",
       " '/home/blake/synergy_transfer/synergy_transfer_data/models/id_feat_gene.p',\n",
       " '/home/blake/synergy_transfer/synergy_transfer_data/summary/processed/codes.p',\n",
       " '/home/blake/synergy_transfer/synergy_transfer_data/summary/processed/summary_mean.p',\n",
       " '/home/blake/synergy_transfer/synergy_transfer_data/summary/raw/summary_table.csv']"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "gdown.download_folder('https://drive.google.com/drive/u/1/folders/14xZ4BLGMEO0Wv5uY0R2tMtB13hevjVLF', quiet=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### <a id='toc1_4_2_4_'></a>[Summary](#toc0_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The first dataset we use is the `summary_table` dataset. This dataset serves as a source for drug and cell_lines information, as well as ids corresponding to various healthcare data sources for joining this information with supplementary data, as we will see later."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "First, load in the summary"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/tmp/ipykernel_147898/107560680.py:1: DtypeWarning: Columns (2,5) have mixed types. Specify dtype option on import or set low_memory=False.\n",
      "  summary = pd.read_csv(f'{DATA_PATH}/summary/raw/summary_table.csv')\n"
     ]
    }
   ],
   "source": [
    "summary = pd.read_csv(f'{DATA_PATH}/summary/raw/summary_table.csv')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Take the average of replicates experiments"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "summary_mean = summary.groupby(['drug_row','drug_col','cell_line_name', 'study_id'], \n",
    "                               as_index=False).mean(numeric_only=True)\n",
    "\n",
    "summary_mean = summary_mean.loc[:, ['drug_row', 'drug_col', 'cell_line_name', 'study_id','ri_row', 'ri_col', 'synergy_loewe']]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### <a id='toc1_4_2_4_1_'></a>[Data Exploration](#toc0_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Determine the Monotherapy Sensitivity"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "sensitivity = pd.concat([\n",
    "          summary_mean.loc[:, ['drug_row', 'cell_line_name', 'ri_row']].rename(columns={'drug_row':'drug', 'ri_row':'ri'}),\n",
    "          summary_mean.loc[:, ['drug_col', 'cell_line_name', 'ri_col']].rename(columns={'drug_col':'drug', 'ri_col':'ri'})\n",
    "        ])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Plot Relative Inhibition vs Loewe Synergy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0, 0.5, 'Counts')"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.hist(sensitivity['ri'], bins=20, range=[-100,100])\n",
    "plt.xlabel(\"Relative Inhibition (Monotherapy Sensitivity)\")\n",
    "plt.ylabel(\"Counts\")\n",
    "#plt.savefig(data_path+'ri_cnt.png',  bbox_inches='tight', dpi=200)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0, 0.5, 'Counts')"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.hist(((summary_mean.loc[:,'synergy_loewe'].values.reshape(-1))), range=[-100, 75])\n",
    "plt.xlabel(\"Loewe Synergy\")\n",
    "plt.ylabel(\"Counts\")\n",
    "#plt.savefig(data_path+'syn_cnt.png', bbox_inches='tight', dpi=200)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(-100.0, 100.0)"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "X = summary_mean.loc[:,'ri_row'].to_numpy()\n",
    "Y = summary_mean.loc[:,'ri_col'].to_numpy()\n",
    "Z = summary_mean.loc[:,'synergy_loewe'].to_numpy()\n",
    "\n",
    "fig = plt.figure()\n",
    "ax = fig.add_subplot(projection='3d')\n",
    "ax.plot_trisurf(X,Y,Z)\n",
    "\n",
    "ax.set_xlabel('Drug1 Relative Inhibition')\n",
    "ax.set_ylabel('Drug2 Relative Inhibition')\n",
    "ax.set_zlabel('Synergy')\n",
    "ax.set_xlim(-100, 100)\n",
    "ax.set_ylim(-100, 100)\n",
    "\n",
    "# fig.savefig(data_path+'ri_synergy_3d.png', dpi=200)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### <a id='toc1_4_2_4_2_'></a>[Transforming Items to ID](#toc0_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We leverage the `Mapping` class to map items to indices and update the datasets to ensure consistency across all of the data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "codes = {\n",
    "    'drugs': Mapping(set(summary_mean['drug_row'].unique()).union(set(summary_mean['drug_col'].unique()))),\n",
    "    'cell':Mapping(summary_mean['cell_line_name'].unique())\n",
    "    }"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "summary_mean['drug_col'] = summary_mean['drug_col'].apply(lambda x: codes['drugs'].item2idx[x])\n",
    "summary_mean['drug_row'] = summary_mean['drug_row'].apply(lambda x: codes['drugs'].item2idx[x])\n",
    "summary_mean['cell_line_name'] = summary_mean['cell_line_name'].apply(lambda x: codes['cell'].item2idx[x])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>drug_row</th>\n",
       "      <th>drug_col</th>\n",
       "      <th>cell_line_name</th>\n",
       "      <th>study_id</th>\n",
       "      <th>ri_row</th>\n",
       "      <th>ri_col</th>\n",
       "      <th>synergy_loewe</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>3528</td>\n",
       "      <td>3467</td>\n",
       "      <td>0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>-21.0794</td>\n",
       "      <td>17.392589</td>\n",
       "      <td>4.436431</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>702</td>\n",
       "      <td>2554</td>\n",
       "      <td>1</td>\n",
       "      <td>6.0</td>\n",
       "      <td>-20.0430</td>\n",
       "      <td>25.595000</td>\n",
       "      <td>-44.555935</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>702</td>\n",
       "      <td>2003</td>\n",
       "      <td>1</td>\n",
       "      <td>6.0</td>\n",
       "      <td>-9.7760</td>\n",
       "      <td>29.111000</td>\n",
       "      <td>-37.189720</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2554</td>\n",
       "      <td>2554</td>\n",
       "      <td>2</td>\n",
       "      <td>12.0</td>\n",
       "      <td>6.7970</td>\n",
       "      <td>6.964000</td>\n",
       "      <td>1.283298</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2554</td>\n",
       "      <td>2554</td>\n",
       "      <td>3</td>\n",
       "      <td>12.0</td>\n",
       "      <td>11.5280</td>\n",
       "      <td>7.190000</td>\n",
       "      <td>-3.028745</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   drug_row  drug_col  cell_line_name  study_id   ri_row     ri_col  \\\n",
       "0      3528      3467               0       4.0 -21.0794  17.392589   \n",
       "1       702      2554               1       6.0 -20.0430  25.595000   \n",
       "2       702      2003               1       6.0  -9.7760  29.111000   \n",
       "3      2554      2554               2      12.0   6.7970   6.964000   \n",
       "4      2554      2554               3      12.0  11.5280   7.190000   \n",
       "\n",
       "   synergy_loewe  \n",
       "0       4.436431  \n",
       "1     -44.555935  \n",
       "2     -37.189720  \n",
       "3       1.283298  \n",
       "4      -3.028745  "
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "summary_mean.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Save the updated summary"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "summary_mean.to_pickle(f'{DATA_PATH}/summary/processed/summary_mean.p')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### <a id='toc1_4_2_5_'></a>[Drugs](#toc0_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Next, we look at the data for `drugs`. This data includes various drug features and molecular structures to learn the model and help it make synergy predictions."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Load in the data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "drug = pd.read_csv(f'{DATA_PATH}/drug/raw/drugs.csv')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### <a id='toc1_4_2_5_1_'></a>[Preprocessing](#toc0_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We apply our mapping and perform some error correction for certain smiles in the dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "drug['id']=drug['dname'].apply(lambda x: codes['drugs'].item2idx.get(x))\n",
    "drug = drug.loc[~drug['id'].isna(),:]\n",
    "\n",
    "drug.loc[drug['cid']==57519530,'smiles']='CC(C)CC1C(=O)NC(C(=O)NC(CCCC=CCCCCCCC(C(=O)NC(C(=O)NC(C(=O)NC(C(=O)NC(C(=O)N1)CCCNC(=N)N)CC2=CNC3=CC=CC=C32)CC(C)C)CC(=O)N)(C)NC(=O)C(CC4=CC=CC=C4)NC(=O)C(C(C)O)NC(=O)C(CCC(=O)N)NC(=O)C(CCC(=O)N)NC(=O)C(CO)NC(=O)C(CCC(=O)N)NC(=O)CCNC(=O)C)(C)C(=O)NC(CCC(=O)N)C(=O)NC(CC(=O)N)C(=O)N)CC(C)C'\n",
    "drug.loc[drug['cid']==73265323,'smiles']='CC(C)CCCCCCCC(=O)NC1C(C(C(OC1OC2=C3C=C4C=C2OC5=C(C=C(C=C5)C(C6C(=O)NC(C7=C(C(=CC(=C7)O)OC8C(C(C(C(O8)CO)O)O)O)C9=C(C=CC(=C9)C(C(=O)N6)NC(=O)C4NC(=O)C1C2=CC(=CC(=C2)OC2=C(C=CC(=C2)C(C(=O)NC(CC2=CC(=C(O3)C=C2)Cl)C(=O)N1)N)O)O)O)C(=O)O)OC1C(C(C(C(O1)CO)O)O)NC(=O)C)Cl)CO)O)O'\n",
    "drug.loc[drug['cid']==16131923,'smiles']='CCCCCCCCCC(=O)NC1C(C(C(OC1OC2=C3C=C4C=C2OC5=C(C=C(C=C5)C(C6C(=O)NC(C7=C(C(=CC(=C7)O)OC8C(C(C(C(O8)CO)O)O)O)C9=C(C=CC(=C9)C(C(=O)N6)NC(=O)C4NC(=O)C1C2=CC(=CC(=C2)OC2=C(C=CC(=C2)C(C(=O)NC(CC2=CC(=C(O3)C=C2)Cl)C(=O)N1)N)O)O)O)C(=O)O)OC1C(C(C(C(O1)CO)O)O)NC(=O)C)Cl)CO)O)O'\n",
    "\n",
    "drug.to_csv(f'{DATA_PATH}/drug/processed/drug_clean.csv', index=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Get the MAACS fingerprint from the smiles for numeric representations"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "drug = pd.read_csv(f'{DATA_PATH}/drug/processed/drug_clean.csv')\n",
    "drug['fps']=drug['smiles'].apply(lambda x: list(MACCSkeys.GenMACCSKeys(Chem.MolFromSmiles(x))) if Chem.MolFromSmiles(x) is not None else '')\n",
    "\n",
    "drug.to_pickle(f'{DATA_PATH}/drug/processed/drug_clean_fp.p')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "drug = pickle.load(open(f'{DATA_PATH}/drug/processed/drug_clean_fp.p', 'rb'))\n",
    "drug['id'] = drug['id'].apply(lambda x: int(x))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>dname</th>\n",
       "      <th>chembl_id</th>\n",
       "      <th>inchikey</th>\n",
       "      <th>smiles</th>\n",
       "      <th>cid</th>\n",
       "      <th>molecular_formula</th>\n",
       "      <th>clinical_phase</th>\n",
       "      <th>cid_m</th>\n",
       "      <th>cid_s</th>\n",
       "      <th>stitch_name</th>\n",
       "      <th>drugbank_id</th>\n",
       "      <th>kegg_id</th>\n",
       "      <th>id</th>\n",
       "      <th>fps</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>VALPROIC ACID</td>\n",
       "      <td>CHEMBL109</td>\n",
       "      <td>NIJJYAXOARWZEE-UHFFFAOYSA-N</td>\n",
       "      <td>CCCC(CCC)C(=O)O</td>\n",
       "      <td>3121</td>\n",
       "      <td>C8H16O2</td>\n",
       "      <td>4</td>\n",
       "      <td>CIDm00003121</td>\n",
       "      <td>CIDs00003121</td>\n",
       "      <td>valproate</td>\n",
       "      <td>DB00313</td>\n",
       "      <td>D00399</td>\n",
       "      <td>1716</td>\n",
       "      <td>[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>PROBUCOL</td>\n",
       "      <td>CHEMBL608</td>\n",
       "      <td>FYPMFJGVHOHGLL-UHFFFAOYSA-N</td>\n",
       "      <td>CC(C)(C)C1=CC(=CC(=C1O)C(C)(C)C)SC(C)(C)SC2=CC...</td>\n",
       "      <td>4912</td>\n",
       "      <td>C31H48O2S2</td>\n",
       "      <td>4</td>\n",
       "      <td>CIDm00004912</td>\n",
       "      <td>CIDs00004912</td>\n",
       "      <td>probucol</td>\n",
       "      <td>DB01599</td>\n",
       "      <td>D00476</td>\n",
       "      <td>1407</td>\n",
       "      <td>[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>CARBAMAZEPINE</td>\n",
       "      <td>CHEMBL108</td>\n",
       "      <td>FFGPTBGBLSHEPO-UHFFFAOYSA-N</td>\n",
       "      <td>C1=CC=C2C(=C1)C=CC3=CC=CC=C3N2C(=O)N</td>\n",
       "      <td>2554</td>\n",
       "      <td>C15H12N2O</td>\n",
       "      <td>4</td>\n",
       "      <td>CIDm00002554</td>\n",
       "      <td>CIDs00002554</td>\n",
       "      <td>carbamazepine</td>\n",
       "      <td>DB00564</td>\n",
       "      <td>D00252</td>\n",
       "      <td>695</td>\n",
       "      <td>[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>PHENYTOIN</td>\n",
       "      <td>CHEMBL16</td>\n",
       "      <td>CXOFVDLJLONNDW-UHFFFAOYSA-N</td>\n",
       "      <td>C1=CC=C(C=C1)C2(C(=O)NC(=O)N2)C3=CC=CC=C3</td>\n",
       "      <td>1775</td>\n",
       "      <td>C15H12N2O2</td>\n",
       "      <td>4</td>\n",
       "      <td>CIDm00001775</td>\n",
       "      <td>CIDs00001775</td>\n",
       "      <td>phenytoin</td>\n",
       "      <td>DB00252</td>\n",
       "      <td>D00512</td>\n",
       "      <td>3844</td>\n",
       "      <td>[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Tesmilifene</td>\n",
       "      <td>CHEMBL26424</td>\n",
       "      <td>NFIXBCVWIPOYCD-UHFFFAOYSA-N</td>\n",
       "      <td>CCN(CC)CCOC1=CC=C(C=C1)CC2=CC=CC=C2</td>\n",
       "      <td>108092</td>\n",
       "      <td>C19H25NO</td>\n",
       "      <td>2</td>\n",
       "      <td>CIDm00108092</td>\n",
       "      <td>CIDs00108092</td>\n",
       "      <td>tesmilifene</td>\n",
       "      <td>DB04905</td>\n",
       "      <td>NaN</td>\n",
       "      <td>515</td>\n",
       "      <td>[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>NOBILETIN</td>\n",
       "      <td>CHEMBL76447</td>\n",
       "      <td>MRIAQLRQZPPODS-UHFFFAOYSA-N</td>\n",
       "      <td>COC1=C(C=C(C=C1)C2=CC(=O)C3=C(O2)C(=C(C(=C3OC)...</td>\n",
       "      <td>72344</td>\n",
       "      <td>C21H22O8</td>\n",
       "      <td>0</td>\n",
       "      <td>CIDm00072344</td>\n",
       "      <td>CIDs00072344</td>\n",
       "      <td>nobiletin</td>\n",
       "      <td>NaN</td>\n",
       "      <td>C10112</td>\n",
       "      <td>1155</td>\n",
       "      <td>[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>(+/-)-SULFINPYRAZONE</td>\n",
       "      <td>CHEMBL832</td>\n",
       "      <td>MBGGBVCUIVRRBF-UHFFFAOYSA-N</td>\n",
       "      <td>C1=CC=C(C=C1)N2C(=O)C(C(=O)N2C3=CC=CC=C3)CCS(=...</td>\n",
       "      <td>5342</td>\n",
       "      <td>C23H20N2O3S</td>\n",
       "      <td>4</td>\n",
       "      <td>CIDm00005342</td>\n",
       "      <td>CIDs00005342</td>\n",
       "      <td>sulfinpyrazone</td>\n",
       "      <td>DB01138</td>\n",
       "      <td>D00449</td>\n",
       "      <td>1355</td>\n",
       "      <td>[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>BENZBROMARONE</td>\n",
       "      <td>CHEMBL388590</td>\n",
       "      <td>WHQCHUCQKNIQEC-UHFFFAOYSA-N</td>\n",
       "      <td>CCC1=C(C2=CC=CC=C2O1)C(=O)C3=CC(=C(C(=C3)Br)O)Br</td>\n",
       "      <td>2333</td>\n",
       "      <td>C17H12Br2O3</td>\n",
       "      <td>4</td>\n",
       "      <td>CIDm00002333</td>\n",
       "      <td>CIDs00002333</td>\n",
       "      <td>benzbromarone</td>\n",
       "      <td>DB12319</td>\n",
       "      <td>D01056</td>\n",
       "      <td>3887</td>\n",
       "      <td>[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>TOLBUTAMIDE</td>\n",
       "      <td>CHEMBL782</td>\n",
       "      <td>JLRGJRBPOGGCBT-UHFFFAOYSA-N</td>\n",
       "      <td>CCCCNC(=O)NS(=O)(=O)C1=CC=C(C=C1)C</td>\n",
       "      <td>5505</td>\n",
       "      <td>C12H18N2O3S</td>\n",
       "      <td>4</td>\n",
       "      <td>CIDm00005505</td>\n",
       "      <td>CIDs00005505</td>\n",
       "      <td>tolbutamide</td>\n",
       "      <td>DB01124</td>\n",
       "      <td>D00380</td>\n",
       "      <td>3785</td>\n",
       "      <td>[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>REPAGLINIDE</td>\n",
       "      <td>CHEMBL1272</td>\n",
       "      <td>FAEKWTJYAYMJKF-QHCPKHFHSA-N</td>\n",
       "      <td>CCOC1=C(C=CC(=C1)CC(=O)NC(CC(C)C)C2=CC=CC=C2N3...</td>\n",
       "      <td>65981</td>\n",
       "      <td>C27H36N2O4</td>\n",
       "      <td>4</td>\n",
       "      <td>CIDm00004547</td>\n",
       "      <td>CIDs00065981</td>\n",
       "      <td>repaglinide</td>\n",
       "      <td>DB00912</td>\n",
       "      <td>D00594</td>\n",
       "      <td>2523</td>\n",
       "      <td>[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, ...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                  dname     chembl_id                     inchikey  \\\n",
       "0         VALPROIC ACID     CHEMBL109  NIJJYAXOARWZEE-UHFFFAOYSA-N   \n",
       "1              PROBUCOL     CHEMBL608  FYPMFJGVHOHGLL-UHFFFAOYSA-N   \n",
       "2         CARBAMAZEPINE     CHEMBL108  FFGPTBGBLSHEPO-UHFFFAOYSA-N   \n",
       "3             PHENYTOIN      CHEMBL16  CXOFVDLJLONNDW-UHFFFAOYSA-N   \n",
       "4           Tesmilifene   CHEMBL26424  NFIXBCVWIPOYCD-UHFFFAOYSA-N   \n",
       "5             NOBILETIN   CHEMBL76447  MRIAQLRQZPPODS-UHFFFAOYSA-N   \n",
       "6  (+/-)-SULFINPYRAZONE     CHEMBL832  MBGGBVCUIVRRBF-UHFFFAOYSA-N   \n",
       "7         BENZBROMARONE  CHEMBL388590  WHQCHUCQKNIQEC-UHFFFAOYSA-N   \n",
       "8           TOLBUTAMIDE     CHEMBL782  JLRGJRBPOGGCBT-UHFFFAOYSA-N   \n",
       "9           REPAGLINIDE    CHEMBL1272  FAEKWTJYAYMJKF-QHCPKHFHSA-N   \n",
       "\n",
       "                                              smiles     cid  \\\n",
       "0                                    CCCC(CCC)C(=O)O    3121   \n",
       "1  CC(C)(C)C1=CC(=CC(=C1O)C(C)(C)C)SC(C)(C)SC2=CC...    4912   \n",
       "2               C1=CC=C2C(=C1)C=CC3=CC=CC=C3N2C(=O)N    2554   \n",
       "3          C1=CC=C(C=C1)C2(C(=O)NC(=O)N2)C3=CC=CC=C3    1775   \n",
       "4                CCN(CC)CCOC1=CC=C(C=C1)CC2=CC=CC=C2  108092   \n",
       "5  COC1=C(C=C(C=C1)C2=CC(=O)C3=C(O2)C(=C(C(=C3OC)...   72344   \n",
       "6  C1=CC=C(C=C1)N2C(=O)C(C(=O)N2C3=CC=CC=C3)CCS(=...    5342   \n",
       "7   CCC1=C(C2=CC=CC=C2O1)C(=O)C3=CC(=C(C(=C3)Br)O)Br    2333   \n",
       "8                 CCCCNC(=O)NS(=O)(=O)C1=CC=C(C=C1)C    5505   \n",
       "9  CCOC1=C(C=CC(=C1)CC(=O)NC(CC(C)C)C2=CC=CC=C2N3...   65981   \n",
       "\n",
       "  molecular_formula  clinical_phase         cid_m         cid_s  \\\n",
       "0           C8H16O2               4  CIDm00003121  CIDs00003121   \n",
       "1        C31H48O2S2               4  CIDm00004912  CIDs00004912   \n",
       "2         C15H12N2O               4  CIDm00002554  CIDs00002554   \n",
       "3        C15H12N2O2               4  CIDm00001775  CIDs00001775   \n",
       "4          C19H25NO               2  CIDm00108092  CIDs00108092   \n",
       "5          C21H22O8               0  CIDm00072344  CIDs00072344   \n",
       "6       C23H20N2O3S               4  CIDm00005342  CIDs00005342   \n",
       "7       C17H12Br2O3               4  CIDm00002333  CIDs00002333   \n",
       "8       C12H18N2O3S               4  CIDm00005505  CIDs00005505   \n",
       "9        C27H36N2O4               4  CIDm00004547  CIDs00065981   \n",
       "\n",
       "      stitch_name drugbank_id kegg_id    id  \\\n",
       "0       valproate     DB00313  D00399  1716   \n",
       "1        probucol     DB01599  D00476  1407   \n",
       "2   carbamazepine     DB00564  D00252   695   \n",
       "3       phenytoin     DB00252  D00512  3844   \n",
       "4     tesmilifene     DB04905     NaN   515   \n",
       "5       nobiletin         NaN  C10112  1155   \n",
       "6  sulfinpyrazone     DB01138  D00449  1355   \n",
       "7   benzbromarone     DB12319  D01056  3887   \n",
       "8     tolbutamide     DB01124  D00380  3785   \n",
       "9     repaglinide     DB00912  D00594  2523   \n",
       "\n",
       "                                                 fps  \n",
       "0  [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, ...  \n",
       "1  [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, ...  \n",
       "2  [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, ...  \n",
       "3  [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, ...  \n",
       "4  [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, ...  \n",
       "5  [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, ...  \n",
       "6  [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, ...  \n",
       "7  [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, ...  \n",
       "8  [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, ...  \n",
       "9  [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, ...  "
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "drug.head(10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Then, we convert the SMILES sequences to indices for zero embedding"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "seqs = drug['smiles'].to_list()\n",
    "chars = set([char for seq in seqs for char in seq])\n",
    "chars = ['']+list(chars) #for zero embedding\n",
    "codes['mole'] = Mapping(chars)\n",
    "drug['smiles'] = drug['smiles'].apply(lambda x : [codes['mole'].item2idx[char] for char in x])\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### <a id='toc1_4_2_5_2_'></a>[Aligning Drug's Target Gene(s)](#toc0_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "After preprocessing the drugs data, we need to align the drugs with their target genes from DrugBank, TTD, and LINC"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Merge target genes from DrugBank"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "drug_DrugBank_target = pd.read_csv(f'{DATA_PATH}/drug/raw/drug_DrugBank_target.csv')\n",
    "drug_DrugBank_target =  drug_DrugBank_target[['Gene', 'drugbank_id']].groupby('drugbank_id').agg(lambda x: set(x)).applymap(list).reset_index()\n",
    "drug = pd.merge(drug, drug_DrugBank_target, how='left', on='drugbank_id')\n",
    "drug.rename(columns={'Gene':'gene_drugbank'}, inplace=True)\n",
    "drug['gene_drugbank'] = drug['gene_drugbank'].apply(lambda x: [] if type(x) is float else x)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Merge target genes from TTD"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "drug_TTD_target=pd.read_csv(f'{DATA_PATH}/drug/raw/drug_TTD_target.csv')\n",
    "drug_TTD_target['TTD_TARGETS']=drug_TTD_target['TTD_TARGETS'].apply(lambda x: re.split(',|;', x))\n",
    "drug=pd.merge(drug, drug_TTD_target.loc[:,['TTD_TARGETS', 'cid']],how='left', on='cid')\n",
    "drug.rename(columns={'TTD_TARGETS':'gene_ttd'}, inplace=True)\n",
    "drug['gene_ttd']=drug['gene_ttd'].apply(lambda x: [] if type(x) is float else x)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Merge target genes from LINC"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "drug_lincs_target = pd.read_csv(f'{DATA_PATH}/drug/raw/drug_lincs_target.csv')\n",
    "drug_lincs_target['target_genes']=drug_lincs_target['target_genes'].apply(lambda x: x.split(',') if type(x) is not float else [])\n",
    "drug = pd.merge(drug, drug_lincs_target.loc[:, ['target_genes', 'cid']], how='left', on='cid')\n",
    "drug.rename(columns={'target_genes':'gene_linc'}, inplace=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Then, we combine all of the target genes identified for a particular drug into a single list"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "drug['gene'] = drug.apply(lambda row: list(set(row['gene_drugbank']+row['gene_ttd']+row['gene_linc'])), axis=1)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "And map the genes accordingly"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "unique_genes_drug = [l for lst in drug['gene'].aggregate(list).values for l in lst]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### <a id='toc1_4_2_6_'></a>[Cells](#toc0_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### <a id='toc1_4_2_6_1_'></a>[Preprocessing](#toc0_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Load in the data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "cell_line = pd.read_csv(f'{DATA_PATH}/cell/raw/cell_lines.csv')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Correct errors in raw data and apply mapping"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "cell_line.replace({'large_intestine':'colon'}, inplace=True)\n",
    "cell_line.drop('id',axis=1, inplace=True)\n",
    "cell_line['cell_id'] = cell_line['name'].apply(lambda x: codes['cell'].item2idx.get(x))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Then, we create a subset of cell line features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "cell_line_tissue = cell_line.loc[~cell_line['cell_id'].isna(), ['cell_id', 'tissue_name', 'disease_name']]\n",
    "cell_line_tissue['cell_id'] = cell_line_tissue['cell_id'].astype(int)\n",
    "cell_line_tissue = cell_line_tissue.groupby('cell_id').first()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "And apply our mapping logic again to map tissue ids to names"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "codes['tissue'] = Mapping(cell_line_tissue['tissue_name'].unique())\n",
    "cell_line_tissue['tissue_name'] = cell_line_tissue['tissue_name'].apply(lambda x: codes['tissue'].item2idx[x])\n",
    "cell_line_tissue.rename(columns={'tissue_name':'tissue_id'}, inplace=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We do the same mapping process for diseases"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "codes['disease'] = Mapping(cell_line_tissue['disease_name'].apply(lambda x: x.split(' ')[-1].lower() if type(x) is not float else '').unique())\n",
    "cell_line_tissue['disease_name'] = cell_line_tissue['disease_name'].apply(lambda x: codes['disease'].item2idx[x.split(' ')[-1].lower()] if type(x) is not float else codes['disease'].item2idx[''])\n",
    "cell_line_tissue.rename(columns={'disease_name':'disease_id'},inplace=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "And perform some more error correction"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "codes['tissue'].item2idx['haematopoietic_and_lymphoid'] = 1\n",
    "codes['tissue'].idx2item[1] = 'hem&lymp'"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now we check out what our mapping logic has done to our tissues and diseases for verification"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'brain': 0,\n",
       " 'haematopoietic_and_lymphoid': 1,\n",
       " 'soft_tissue': 2,\n",
       " 'bone': 3,\n",
       " 'kidney': 4,\n",
       " 'lung': 5,\n",
       " 'breast': 6,\n",
       " 'colon': 7,\n",
       " 'prostate': 8,\n",
       " 'ovary': 9,\n",
       " 'skin': 10,\n",
       " 'endometrium': 11,\n",
       " 'pancreas': 12}"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "codes['tissue'].item2idx"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'c3058': 0,\n",
       " 'c8851': 1,\n",
       " 'c8971': 2,\n",
       " 'c94764': 3,\n",
       " 'c3716': 4,\n",
       " 'c9357': 5,\n",
       " 'c4817': 6,\n",
       " 'c3174': 7,\n",
       " 'c9385': 8,\n",
       " 'c3512': 9,\n",
       " 'c6975': 10,\n",
       " 'c4194': 11,\n",
       " 'c4033': 12,\n",
       " 'c7953': 13,\n",
       " 'c4349': 14,\n",
       " 'c4863': 15,\n",
       " 'c2926': 16,\n",
       " 'c4910': 17,\n",
       " 'c9154': 18,\n",
       " 'c7979': 19,\n",
       " 'c3802': 20,\n",
       " 'c3224': 21,\n",
       " 'c5214': 22,\n",
       " 'c9142': 23,\n",
       " 'c3493': 24,\n",
       " 'c2923': 25,\n",
       " 'c4450': 26,\n",
       " 'c105555': 27,\n",
       " 'c3242': 28,\n",
       " 'c60781': 29,\n",
       " 'c3796': 30,\n",
       " 'c3510': 31,\n",
       " 'c7978': 32,\n",
       " 'c7873': 33,\n",
       " 'c27677': 34,\n",
       " 'c4872': 35,\n",
       " 'c40078': 36,\n",
       " 'c9351': 37,\n",
       " 'c7700': 38,\n",
       " 'c4908': 39,\n",
       " 'c9383': 40,\n",
       " 'c9120': 41}"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "codes['disease'].item2idx"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We count the number of cell lines per tissue"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0, 0.5, 'Number of Cell Lines')"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 800x500 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "tissue_cell_cnts = cell_line_tissue.groupby('tissue_id').count().sort_values(by='disease_id', ascending=False)['disease_id']\n",
    "plt.figure(figsize=(8,5))\n",
    "plt.bar([codes['tissue'].idx2item[tissue_id] for tissue_id in tissue_cell_cnts.index], tissue_cell_cnts)\n",
    "plt.xticks(rotation=90)\n",
    "plt.xlabel('Tissue')\n",
    "plt.ylabel('Number of Cell Lines')\n",
    "\n",
    "#plt.savefig(data_path+'tissue_num_cell.png', bbox_inches='tight', dpi=200)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### <a id='toc1_4_2_6_2_'></a>[Aligning Cell Lines with Gene Expression(s)](#toc0_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Next, using the `gene_identifiers` and `cell_line_meta` files, we can align the cell lines with their respective gene expressions from Broad Institute and SANGAR"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Load in the data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "gene_identifiers = pd.read_csv(f'{DATA_PATH}/gene/raw/gene_identifiers_latest.csv')\n",
    "cell_line_meta = pd.read_csv(f'{DATA_PATH}/cell/raw/cell_line_meta.csv')\n",
    "fpkm_broad = pd.read_csv(f'{DATA_PATH}/gene/raw/fpkm_broad.csv')\n",
    "fpkm_sangar = pd.read_csv(f'{DATA_PATH}/gene/raw/fpkm_sangar.csv')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Map the gene ids to the cell lines"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [],
   "source": [
    "_codes={}\n",
    "_codes['gene_id2hgnc_symbol'] = gene_identifiers[['gene_id', 'hgnc_symbol']].set_index('gene_id').to_dict('index')\n",
    "_codes['SANGAR_ID2cell_name'] = cell_line_meta.loc[cell_line_meta['name'].isin(codes['cell'].idx2item) & ~cell_line_meta['SANGAR_ID'].isna(), ['SANGAR_ID', 'name']].set_index('SANGAR_ID').to_dict('index')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Align the gene expressions from Broad Institute"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [],
   "source": [
    "fpkm_broad = fpkm_broad.loc[fpkm_broad['model_id'].isin(_codes['SANGAR_ID2cell_name'])]#apply(lambda x: _codes['SANGAR_ID2cell_name'].get(x))\n",
    "fpkm_broad['model_id'] = fpkm_broad['model_id'].apply(lambda x: codes['cell'].item2idx[_codes['SANGAR_ID2cell_name'][x]['name']])\n",
    "fpkm_broad.rename(columns={col:_codes['gene_id2hgnc_symbol'][col]['hgnc_symbol'] for col in fpkm_broad.columns[1:]}, inplace=True)\n",
    "\n",
    "fpkm_broad.rename(columns={'model_id':'cell_id'}, inplace=True)\n",
    "fpkm_broad.set_index('cell_id', inplace=True)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Sanity check"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "21732"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "(fpkm_broad.sum()>0).sum()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Align gene expressions from SANGAR"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [],
   "source": [
    "fpkm_sangar = pd.read_csv(f'{DATA_PATH}/gene/raw/fpkm_sangar.csv')\n",
    "\n",
    "fpkm_sangar = fpkm_sangar.loc[fpkm_sangar['model_id'].isin(_codes['SANGAR_ID2cell_name'])]#apply(lambda x: _codes['SANGAR_ID2cell_name'].get(x))\n",
    "fpkm_sangar['model_id'] = fpkm_sangar['model_id'].apply(lambda x: codes['cell'].item2idx[_codes['SANGAR_ID2cell_name'][x]['name']])\n",
    "fpkm_sangar.rename(columns={col:_codes['gene_id2hgnc_symbol'][col]['hgnc_symbol'] for col in fpkm_sangar.columns[1:]}, inplace=True)\n",
    "\n",
    "fpkm_sangar.rename(columns={'model_id':'cell_id'}, inplace=True)\n",
    "fpkm_sangar.set_index('cell_id', inplace=True)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Sanity check"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "18421"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "(fpkm_sangar.sum()>0).sum()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Finally, integrate gene expressions from Broad Institute and SANGAR into the cell lines"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [],
   "source": [
    "cell_gene_fpkm = pd.concat([fpkm_broad,fpkm_sangar], sort=True) \n",
    "cell_gene_fpkm = cell_gene_fpkm.loc[:,(cell_gene_fpkm.std(skipna=True)!=0)] #drop genes with std=0\n",
    "cell_gene_fpkm.fillna(0, inplace=True)\n",
    "\n",
    "cell_gene_fpkm_norm = ((cell_gene_fpkm-cell_gene_fpkm.mean())/cell_gene_fpkm.std()) #z-score, gene-wise\n",
    "\n",
    "unique_genes_cell = cell_gene_fpkm_norm.columns.to_list()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Check the unique genes across all cell lines"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "22586"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(unique_genes_cell)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### <a id='toc1_4_2_6_3_'></a>[Gene Integration](#toc0_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "After getting the unique genes for the cell lines, we can map them to indices in our codes "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [],
   "source": [
    "codes['gene'] = Mapping(set(unique_genes_cell).union(set(unique_genes_drug)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "And again count the number of unique genes we have mapped"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "24342"
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(codes['gene'].idx2item)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Finally, we map the genes back to our drugs data and extract a subset of drug information to serve as our features for training the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [],
   "source": [
    "drug['gene_id']=drug['gene'].apply(lambda genes: [codes['gene'].item2idx[gene] for gene in genes])\n",
    "drug_features = drug[['id', 'smiles', 'fps', 'gene_id']].groupby('id').first()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Save the drug features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [],
   "source": [
    "drug_features.to_pickle(f'{DATA_PATH}/drug/processed/drug_features.p')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We then merge the cell line's tissue and disease type with the gene"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [],
   "source": [
    "cell_gene = cell_gene_fpkm_norm.apply(lambda row: {codes['gene'].item2idx[gene]:row[gene] for gene in cell_gene_fpkm_norm.columns}, axis=1).reset_index()\n",
    "cell_gene.rename(columns={0:'gene_id'}, inplace=True)\n",
    "cell_gene = cell_gene.groupby('cell_id').first()\n",
    "cell_line = pd.merge(cell_line_tissue, cell_gene, how='left', on='cell_id')\n",
    "cell_line['gene_id']=cell_line['gene_id'].apply(lambda x: {} if type(x) is float else x)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "And save the updated cell lines and our code map for later use"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [],
   "source": [
    "cell_line.to_pickle(f'{DATA_PATH}/cell/processed/cell_features.p')\n",
    "pickle.dump(codes, open(f'{DATA_PATH}/summary/processed/codes.p', 'wb'))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### <a id='toc1_4_2_6_4_'></a>[Data Exploration](#toc0_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "For data exploration and visualization purposes, we will add the cell line's tissue information into our summary table"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [],
   "source": [
    "summary_mean['tissue'] = summary_mean['cell_line_name'].apply(lambda x: cell_line_tissue.loc[x, 'tissue_id'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Plot number of blocks per cell line"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0, 0.5, 'number of blocks')"
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 2000x600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "cell_line_exp = summary_mean.groupby('cell_line_name').count()['drug_row'].sort_values(ascending=False)\n",
    "\n",
    "plt.figure(figsize=(20,6))\n",
    "plt.bar([codes['cell'].idx2item[cell_id]\n",
    "         +'('+codes['tissue'].idx2item[cell_line_tissue.loc[cell_id, 'tissue_id']]\n",
    "         +')' for cell_id in cell_line_exp.index], cell_line_exp)\n",
    "plt.xticks(rotation=90)\n",
    "plt.xlabel('cell line')\n",
    "plt.ylabel('number of blocks')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Plot number of blocks per tissue\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0, 0.5, 'number of drug combinations')"
      ]
     },
     "execution_count": 50,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "tissue_exp = summary_mean.groupby('tissue').count()['drug_row'].sort_values(ascending=False)\n",
    "\n",
    "plt.bar([codes['tissue'].idx2item[tissue_id] for tissue_id in tissue_exp.index], tissue_exp)\n",
    "plt.xticks(rotation=90)\n",
    "plt.xlabel('tissue')\n",
    "plt.ylabel('number of drug combinations')\n",
    "#plt.savefig(data_path+'tissue_num_comb.png', bbox_inches='tight', dpi=200)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Plot number of blocks per study & tissue"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [],
   "source": [
    "study_ids = ['', 'ONEIL','CLOUD','ALMANAC','FORCINA','NCATS_ATL','Mathews','NCATS_DIPG','NCATS_ES(FAKi/AURKi)','NCATS_ES(Nampt+PARP)','Wilson','NCATS_HL','Yohe','NCATS_2D_3D','Phelan','NCATS_MDR_CS','CCLE','CTRPv2','FIMM','gCSI','GDSC1','GRAY','UHNBreast','BEATAML'] #from study.csv\n",
    "num_blocks = summary_mean.groupby(['study_id', 'tissue'], as_index=False)['drug_row'].count().rename(columns={'drug_row':'num_blocks'})\n",
    "num_blocks['tissue'] = num_blocks['tissue'].apply(lambda x: codes['tissue'].idx2item[x])\n",
    "num_blocks['study_id'] = num_blocks['study_id'].apply(lambda x: study_ids[int(x)])\n",
    "\n",
    "# num_blocks_pivot=num_blocks.pivot(\"study_id\", \"tissue\", \"num_blocks\").fillna(0)\n",
    "num_blocks_pivot = num_blocks.pivot(index=\"study_id\", columns=\"tissue\", values=\"num_blocks\").fillna(0)\n",
    "num_blocks_pivot = num_blocks_pivot.loc[['ALMANAC', 'CLOUD', 'NCATS_HL', 'FORCINA','Mathews','Wilson','Yohe','NCATS_MDR_CS','Phelan','NCATS_ATL','NCATS_2D_3D',\n",
    "                      'NCATS_DIPG', 'NCATS_ES(FAKi/AURKi)',\n",
    "                      'NCATS_ES(Nampt+PARP)','ONEIL'],\n",
    "                     ['brain','breast','colon','endometrium','hem&lymp','kidney','lung','ovary',\n",
    "                      'bone','prostate']]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "log_norm = LogNorm(vmin=num_blocks_pivot.min().min()+1, vmax=num_blocks_pivot.max().max())\n",
    "cbar_ticks = [math.pow(10, i) for i in range(math.floor(math.log10(num_blocks_pivot.min().min()+1)), 1+math.ceil(math.log10(num_blocks_pivot.max().max())))]\n",
    "\n",
    "cmap = sns.cubehelix_palette(light=1, as_cmap=True)\n",
    "\n",
    "ax = sns.heatmap(num_blocks_pivot.replace(0,1),\n",
    "           norm=log_norm,\n",
    "            cbar_kws={'ticks':cbar_ticks},\n",
    "            cmap=cmap)\n",
    "\n",
    "#fig=ax.get_figure()\n",
    "#fig.tight_layout()\n",
    "#fig.savefig(data_path+'study_heatmap.png', dpi=200)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Counts number of unique drugs for each tissue"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [],
   "source": [
    "tissue_drugs = pd.merge(summary_mean.groupby('tissue')['drug_row'].apply(set).reset_index(name='drug_row'),\n",
    "         summary_mean.groupby('tissue')['drug_col'].apply(set).reset_index(name='drug_row'),\n",
    "         on='tissue')\n",
    "tissue_drugs['drugs']=tissue_drugs.apply(lambda row: row[1].union(row[2]), axis=1)\n",
    "tissue_drugs.drop(['drug_row_x','drug_row_y'], axis=1, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [],
   "source": [
    "tissue_drugs['num_drugs']=tissue_drugs['drugs'].apply(lambda x: len(x))\n",
    "tissue_drugs.sort_values('num_drugs', ascending=False, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0, 0.5, 'number of unique drugs')"
      ]
     },
     "execution_count": 55,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.bar([codes['tissue'].idx2item[tissue_id] for tissue_id in tissue_drugs['tissue']], tissue_drugs['num_drugs'])\n",
    "plt.xticks(rotation=90)\n",
    "plt.xlabel('tissue')\n",
    "plt.ylabel('number of unique drugs')\n",
    "#plt.savefig(data_path+'tissue_num_drugs.png', bbox_inches='tight', dpi=200)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Count number of shared drugs between tissues"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[None,\n",
       " None,\n",
       " None,\n",
       " None,\n",
       " None,\n",
       " None,\n",
       " None,\n",
       " None,\n",
       " None,\n",
       " None,\n",
       " None,\n",
       " None,\n",
       " None,\n",
       " None,\n",
       " None,\n",
       " None,\n",
       " None,\n",
       " None,\n",
       " None,\n",
       " None,\n",
       " None,\n",
       " None,\n",
       " None,\n",
       " None,\n",
       " None,\n",
       " None]"
      ]
     },
     "execution_count": 56,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "num_intersects=np.zeros((len(tissue_drugs),len(tissue_drugs)))\n",
    "for i, tissue_id in enumerate(tissue_drugs['tissue']):    \n",
    "    num_intersects[i,:]=tissue_drugs['drugs'].apply(lambda x: len(x.intersection(tissue_drugs.loc[tissue_id, 'drugs']))).values\n",
    "\n",
    "\n",
    "fig,ax=plt.subplots()\n",
    "im=ax.imshow(num_intersects)\n",
    "ax.tick_params(top=True, bottom=False, labeltop=True, labelbottom=False)\n",
    "ax.set_xticks(np.arange(len(tissue_drugs)))\n",
    "ax.set_yticks(np.arange(len(tissue_drugs)))\n",
    "ax.set_xticklabels([codes['tissue'].idx2item[tissue_id] for tissue_id in tissue_drugs['tissue']])\n",
    "ax.set_yticklabels([codes['tissue'].idx2item[tissue_id] for tissue_id in tissue_drugs['tissue']])\n",
    "\n",
    "fig.colorbar(im, ax=ax)\n",
    "\n",
    "plt.setp(ax.get_xticklabels(), rotation=90)\n",
    "#fig.tight_layout()\n",
    "#plt.savefig(data_path+'heatmap.png', bb_inches='tight', dpi=200)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "At this point, we have successfully preprocessed all of the data required to train the model. In the next section, we will load this data back in, define the model architecture, and begin training"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### <a id='toc1_4_3_'></a>[Model](#toc0_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### <a id='toc1_4_3_1_'></a>[Architecture](#toc0_)\n",
    "\n",
    "The model architecture is designed to integrate multi-modal data inputs and predict two main outputs: monotherapy sensitivity and combination drug synergy. This architecture comprises several components:\n",
    "\n",
    "##### <a id='toc1_4_3_1_1_'></a>[Drug Encoder](#toc0_)\n",
    "\n",
    "This component processes the features of each drug, which include drug ID, MACCS fingerprints, canonical SMILES (a text notation for describing the structure of chemical species), and target genes. These features are transformed into embeddings, which are compact vector representations that capture the essence of the input features. SMILES sequences, which vary in length, are handled by a Transformer encoder—a type of model that is highly effective in managing sequence data in natural language processing.\n",
    "\n",
    "<div align=\"center\">\n",
    "    <img src=\"public/assets/drug_encoder_figure.png\" width=\"300\">\n",
    "    <p><b>Figure 1:</b> Drug Encoder Diagram from <a href=\"https://doi.org/10.1101/2020.02.05.932657\">Kim et al. (2020)</a></p>\n",
    "</div>\n",
    "\n",
    "##### <a id='toc1_4_3_1_2_'></a>[Cell Line Encoder](#toc0_)\n",
    "\n",
    "Similar to the drug encoder, this component processes features related to cell lines, such as cell line ID, tissue type, cancer type, and gene expression profiles. These features are also converted into embeddings to facilitate further analysis.\n",
    "\n",
    "<div align=\"center\">\n",
    "    <img src=\"public/assets/cell_line_encoder_figure.png\" width=\"300\">\n",
    "    <p><b>Figure 2:</b> Cell Line Encoder Diagram from <a href=\"https://doi.org/10.1101/2020.02.05.932657\">Kim et al. (2020)</a></p>\n",
    "</div>\n",
    "\n",
    "##### <a id='toc1_4_3_1_3_'></a>[Merging Layers](#toc0_)\n",
    "\n",
    " The embeddings from the drug encoder and the cell line encoder are combined and fed into a neural network to predict the drug response. This part of the network uses feed-forward layers with ReLU activation functions to process the combined embeddings for final output predictions.\n",
    "\n",
    " <div align=\"center\">\n",
    "    <img src=\"public/assets/encoding_scheme_figure.png\" width=\"500\">\n",
    "    <p><b>Figure 3:</b> Merging Drug and Cell Lines Encoders Diagram from <a href=\"https://doi.org/10.1101/2020.02.05.932657\">Kim et al. (2020)</a></p>\n",
    "</div>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### <a id='toc1_4_3_2_'></a>[Configuration](#toc0_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Import modules"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import torch\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import pdb\n",
    "import time\n",
    "import pickle\n",
    "import logging\n",
    "import matplotlib.pyplot as plt\n",
    "from utilities import Mapping\n",
    "\n",
    "\n",
    "import torch.nn as nn\n",
    "from torch.autograd import Variable\n",
    "import torch.optim as optim\n",
    "import torch.nn.functional as F\n",
    "from torch.utils.data import Dataset, DataLoader\n",
    "\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn import metrics "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Set parameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [],
   "source": [
    "DATA_PATH='synergy_transfer_data'\n",
    "\n",
    "bsz=128\n",
    "cuda=True\n",
    "device=0\n",
    "torch.cuda.set_device(device)\n",
    "\n",
    "num_gene_compressed_drug=64\n",
    "num_gene_compressed_cell=128\n",
    "\n",
    "#isClassification=True #False for regression task\n",
    "syn_threshold=30\n",
    "ri_threshold=50\n",
    "\n",
    "log_interval=100\n",
    "epochs=10"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Load the summary and codes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [],
   "source": [
    "df=pickle.load(open(f'{DATA_PATH}/summary/processed/summary_mean.p', 'rb'))\n",
    "codes=pickle.load(open(f'{DATA_PATH}/summary/processed/codes.p', 'rb'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>drug_row</th>\n",
       "      <th>drug_col</th>\n",
       "      <th>cell_line_name</th>\n",
       "      <th>study_id</th>\n",
       "      <th>ri_row</th>\n",
       "      <th>ri_col</th>\n",
       "      <th>synergy_loewe</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>3528</td>\n",
       "      <td>3467</td>\n",
       "      <td>0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>-21.0794</td>\n",
       "      <td>17.392589</td>\n",
       "      <td>4.436431</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>702</td>\n",
       "      <td>2554</td>\n",
       "      <td>1</td>\n",
       "      <td>6.0</td>\n",
       "      <td>-20.0430</td>\n",
       "      <td>25.595000</td>\n",
       "      <td>-44.555935</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>702</td>\n",
       "      <td>2003</td>\n",
       "      <td>1</td>\n",
       "      <td>6.0</td>\n",
       "      <td>-9.7760</td>\n",
       "      <td>29.111000</td>\n",
       "      <td>-37.189720</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2554</td>\n",
       "      <td>2554</td>\n",
       "      <td>2</td>\n",
       "      <td>12.0</td>\n",
       "      <td>6.7970</td>\n",
       "      <td>6.964000</td>\n",
       "      <td>1.283298</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2554</td>\n",
       "      <td>2554</td>\n",
       "      <td>3</td>\n",
       "      <td>12.0</td>\n",
       "      <td>11.5280</td>\n",
       "      <td>7.190000</td>\n",
       "      <td>-3.028745</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   drug_row  drug_col  cell_line_name  study_id   ri_row     ri_col  \\\n",
       "0      3528      3467               0       4.0 -21.0794  17.392589   \n",
       "1       702      2554               1       6.0 -20.0430  25.595000   \n",
       "2       702      2003               1       6.0  -9.7760  29.111000   \n",
       "3      2554      2554               2      12.0   6.7970   6.964000   \n",
       "4      2554      2554               3      12.0  11.5280   7.190000   \n",
       "\n",
       "   synergy_loewe  \n",
       "0       4.436431  \n",
       "1     -44.555935  \n",
       "2     -37.189720  \n",
       "3       1.283298  \n",
       "4      -3.028745  "
      ]
     },
     "execution_count": 60,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Load drug external features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [],
   "source": [
    "drug_features=pickle.load(open(f'{DATA_PATH}/drug/processed/drug_features.p', 'rb'))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Load cell external features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [],
   "source": [
    "cell_features=pickle.load(open(f'{DATA_PATH}/cell/processed/cell_features.p', 'rb'))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Set parameters based on size of data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [],
   "source": [
    "num_celllines= len(codes['cell'].idx2item)\n",
    "num_drugs=len(codes['drugs'].idx2item)\n",
    "\n",
    "num_genes = len(codes['gene'].idx2item)\n",
    "num_tissue = len(codes['tissue'].idx2item)\n",
    "num_disease = len(codes['disease'].idx2item)\n",
    "\n",
    "num_drug_fp=len(drug_features.loc[0,'fps'])\n",
    "max_drug_sm_len = drug_features['smiles'].apply(lambda x: len(x)).max()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### <a id='toc1_4_3_3_'></a>[Gene Compression](#toc0_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### <a id='toc1_4_3_3_1_'></a>[Dataset and Dataloader](#toc0_)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [],
   "source": [
    "class DrugTargetDataset(Dataset):\n",
    "    def __init__(self, drug_features):\n",
    "        self.drug_features = drug_features\n",
    "    def __len__(self):\n",
    "        return len(self.drug_features)\n",
    "    def __getitem__(self,idx):\n",
    "        gene_ids=self.drug_features.loc[idx, 'gene_id']\n",
    "        genes=np.zeros(num_genes)\n",
    "        genes[gene_ids]=1\n",
    "        \n",
    "        return genes\n",
    "    \n",
    "class CellGeneDataset(Dataset):\n",
    "    def __init__(self, cell_features):\n",
    "        self.cell_features = cell_features\n",
    "    def __len__(self):\n",
    "        return len(self.cell_features)\n",
    "    def __getitem__(self,idx):\n",
    "        gene_ids=self.cell_features.loc[idx,'gene_id']\n",
    "        genes = np.zeros(num_genes)\n",
    "        for key,value in gene_ids.items():\n",
    "            genes[key]=value\n",
    "            \n",
    "        return genes\n",
    "            "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Two layers of fully connected layers\n",
    "class FC2(nn.Module):\n",
    "    def __init__(self, in_features, out_features, dropout):\n",
    "        super(FC2, self).__init__()\n",
    "        \n",
    "        self.bn = nn.BatchNorm1d(in_features)\n",
    "        self.fc1 = nn.Linear(in_features, int(in_features/2))\n",
    "        self.fc2 = nn.Linear(int(in_features/2),out_features)\n",
    "        self.dropout= nn.Dropout(dropout)\n",
    "                \n",
    "    def forward(self, x):\n",
    "        x = self.bn(x)\n",
    "        x = self.dropout(x)\n",
    "        x = self.dropout(F.relu(self.fc1(x)))\n",
    "        x = self.fc2(x)\n",
    "        \n",
    "        return x\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### <a id='toc1_4_3_3_2_'></a>[Compress gene features](#toc0_)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [],
   "source": [
    "class GeneCompressor(nn.Module):\n",
    "    def __init__(self, num_in, num_out, dropout=0.1):\n",
    "        super(GeneCompressor, self).__init__()\n",
    "        self.dropout=dropout\n",
    "        self.encoder=nn.Linear(num_in, num_out)\n",
    "        self.decoder=nn.Linear(num_out,num_in)\n",
    "\n",
    "    def _encoder(self,x):\n",
    "        return F.dropout(F.relu(self.encoder(x)), self.dropout, training=self.training)\n",
    "    \n",
    "    def _decoder(self,x):\n",
    "        return F.dropout(self.decoder(x), self.dropout, training=self.training)\n",
    "    \n",
    "    def forward(self, x):\n",
    "        x=self._encoder(x)\n",
    "        x=self._decoder(x)\n",
    "        return x"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [],
   "source": [
    "def geneCompressing(data_loader,num_gene_compressed, noise_weight=0.2, epochs=20, log_interval=10 ):\n",
    "    #model\n",
    "    geneCompressor=GeneCompressor(num_genes, num_out=num_gene_compressed, dropout=0.1)\n",
    "    if cuda: \n",
    "        geneCompressor=geneCompressor.cuda()\n",
    "    criterion=nn.MSELoss()\n",
    "    optimizer=optim.Adam(geneCompressor.parameters())\n",
    "\n",
    "    for epoch in range(1,epochs+1):\n",
    "        #train\n",
    "        geneCompressor.train()\n",
    "        total_loss=0\n",
    "        start_time=time.time()\n",
    "        for iteration, gene in enumerate(data_loader):\n",
    "            gene=Variable(gene).float()\n",
    "            noise=noise_weight*torch.randn(gene.shape)\n",
    "\n",
    "            if cuda:\n",
    "                gene=gene.cuda()\n",
    "                noise=noise.cuda()\n",
    "            optimizer.zero_grad()\n",
    "            output=geneCompressor(gene+noise)\n",
    "            loss=criterion(output,gene)\n",
    "            loss.backward()\n",
    "            optimizer.step()\n",
    "            total_loss += loss.data\n",
    "            if iteration % log_interval == 0 and iteration > 0:\n",
    "                cur_loss = total_loss.item() / log_interval\n",
    "                elapsed = time.time() - start_time\n",
    "                print('| epoch {:3d} | {:5d}/{:5d} batches | ms/batch {:5.2f} | loss {:8.5f}'.format(epoch, iteration, int(len(data_loader)/bsz), elapsed * 1000/log_interval, cur_loss))\n",
    "                total_loss = 0\n",
    "                start_time = time.time()\n",
    "#         #test\n",
    "#         geneCompressor.eval()\n",
    "#         total_loss=0\n",
    "#         start_time=time.time()\n",
    "#         with torch.no_grad():\n",
    "#             for iteration, gene in enumerate(test_data_loader):\n",
    "#                 gene=Variable(gene).float()\n",
    "#                 if cuda:\n",
    "#                     gene=gene.cuda(device)\n",
    "#                 output=geneCompressor(gene)\n",
    "#                 loss=criterion(output,gene)\n",
    "#                 total_loss += loss.data\n",
    "#             print(total_loss.item()/iteration)\n",
    "    return geneCompressor\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "| epoch   1 |    10/    0 batches | ms/batch 31.56 | loss  0.00790\n",
      "| epoch   1 |    20/    0 batches | ms/batch 14.72 | loss  0.00530\n",
      "| epoch   1 |    30/    0 batches | ms/batch 15.43 | loss  0.00506\n",
      "| epoch   1 |    40/    0 batches | ms/batch 15.96 | loss  0.00512\n",
      "| epoch   1 |    50/    0 batches | ms/batch 15.98 | loss  0.00444\n",
      "| epoch   1 |    60/    0 batches | ms/batch 15.23 | loss  0.00451\n",
      "| epoch   2 |    10/    0 batches | ms/batch 16.78 | loss  0.00410\n",
      "| epoch   2 |    20/    0 batches | ms/batch 14.97 | loss  0.00347\n",
      "| epoch   2 |    30/    0 batches | ms/batch 14.86 | loss  0.00289\n",
      "| epoch   2 |    40/    0 batches | ms/batch 16.68 | loss  0.00247\n",
      "| epoch   2 |    50/    0 batches | ms/batch 15.74 | loss  0.00204\n",
      "| epoch   2 |    60/    0 batches | ms/batch 16.15 | loss  0.00181\n",
      "| epoch   3 |    10/    0 batches | ms/batch 18.75 | loss  0.00154\n",
      "| epoch   3 |    20/    0 batches | ms/batch 15.98 | loss  0.00117\n",
      "| epoch   3 |    30/    0 batches | ms/batch 16.10 | loss  0.00101\n",
      "| epoch   3 |    40/    0 batches | ms/batch 15.76 | loss  0.00087\n",
      "| epoch   3 |    50/    0 batches | ms/batch 16.22 | loss  0.00071\n",
      "| epoch   3 |    60/    0 batches | ms/batch 15.93 | loss  0.00063\n",
      "| epoch   4 |    10/    0 batches | ms/batch 17.64 | loss  0.00058\n",
      "| epoch   4 |    20/    0 batches | ms/batch 15.90 | loss  0.00045\n",
      "| epoch   4 |    30/    0 batches | ms/batch 16.27 | loss  0.00042\n",
      "| epoch   4 |    40/    0 batches | ms/batch 16.35 | loss  0.00036\n",
      "| epoch   4 |    50/    0 batches | ms/batch 15.90 | loss  0.00036\n",
      "| epoch   4 |    60/    0 batches | ms/batch 16.49 | loss  0.00029\n",
      "| epoch   5 |    10/    0 batches | ms/batch 17.33 | loss  0.00028\n",
      "| epoch   5 |    20/    0 batches | ms/batch 15.39 | loss  0.00024\n",
      "| epoch   5 |    30/    0 batches | ms/batch 16.12 | loss  0.00021\n",
      "| epoch   5 |    40/    0 batches | ms/batch 16.90 | loss  0.00020\n",
      "| epoch   5 |    50/    0 batches | ms/batch 14.91 | loss  0.00024\n",
      "| epoch   5 |    60/    0 batches | ms/batch 15.80 | loss  0.00018\n",
      "| epoch   6 |    10/    0 batches | ms/batch 16.84 | loss  0.00018\n",
      "| epoch   6 |    20/    0 batches | ms/batch 15.13 | loss  0.00017\n",
      "| epoch   6 |    30/    0 batches | ms/batch 16.14 | loss  0.00020\n",
      "| epoch   6 |    40/    0 batches | ms/batch 15.78 | loss  0.00015\n",
      "| epoch   6 |    50/    0 batches | ms/batch 15.27 | loss  0.00014\n",
      "| epoch   6 |    60/    0 batches | ms/batch 16.05 | loss  0.00014\n",
      "| epoch   7 |    10/    0 batches | ms/batch 16.60 | loss  0.00015\n",
      "| epoch   7 |    20/    0 batches | ms/batch 15.73 | loss  0.00019\n",
      "| epoch   7 |    30/    0 batches | ms/batch 16.48 | loss  0.00012\n",
      "| epoch   7 |    40/    0 batches | ms/batch 16.28 | loss  0.00012\n",
      "| epoch   7 |    50/    0 batches | ms/batch 16.54 | loss  0.00012\n",
      "| epoch   7 |    60/    0 batches | ms/batch 17.42 | loss  0.00013\n",
      "| epoch   8 |    10/    0 batches | ms/batch 17.72 | loss  0.00014\n",
      "| epoch   8 |    20/    0 batches | ms/batch 15.27 | loss  0.00013\n",
      "| epoch   8 |    30/    0 batches | ms/batch 15.66 | loss  0.00013\n",
      "| epoch   8 |    40/    0 batches | ms/batch 15.55 | loss  0.00016\n",
      "| epoch   8 |    50/    0 batches | ms/batch 15.19 | loss  0.00009\n",
      "| epoch   8 |    60/    0 batches | ms/batch 15.02 | loss  0.00011\n",
      "| epoch   9 |    10/    0 batches | ms/batch 16.58 | loss  0.00011\n",
      "| epoch   9 |    20/    0 batches | ms/batch 15.60 | loss  0.00012\n",
      "| epoch   9 |    30/    0 batches | ms/batch 15.75 | loss  0.00017\n",
      "| epoch   9 |    40/    0 batches | ms/batch 15.27 | loss  0.00012\n",
      "| epoch   9 |    50/    0 batches | ms/batch 15.52 | loss  0.00011\n",
      "| epoch   9 |    60/    0 batches | ms/batch 15.47 | loss  0.00011\n",
      "| epoch  10 |    10/    0 batches | ms/batch 18.16 | loss  0.00014\n",
      "| epoch  10 |    20/    0 batches | ms/batch 15.32 | loss  0.00009\n",
      "| epoch  10 |    30/    0 batches | ms/batch 16.52 | loss  0.00011\n",
      "| epoch  10 |    40/    0 batches | ms/batch 16.47 | loss  0.00012\n",
      "| epoch  10 |    50/    0 batches | ms/batch 15.38 | loss  0.00011\n",
      "| epoch  10 |    60/    0 batches | ms/batch 15.63 | loss  0.00017\n",
      "| epoch  11 |    10/    0 batches | ms/batch 17.88 | loss  0.00013\n",
      "| epoch  11 |    20/    0 batches | ms/batch 14.94 | loss  0.00011\n",
      "| epoch  11 |    30/    0 batches | ms/batch 15.12 | loss  0.00012\n",
      "| epoch  11 |    40/    0 batches | ms/batch 16.26 | loss  0.00016\n",
      "| epoch  11 |    50/    0 batches | ms/batch 14.60 | loss  0.00010\n",
      "| epoch  11 |    60/    0 batches | ms/batch 15.19 | loss  0.00011\n",
      "| epoch  12 |    10/    0 batches | ms/batch 18.22 | loss  0.00011\n",
      "| epoch  12 |    20/    0 batches | ms/batch 15.23 | loss  0.00011\n",
      "| epoch  12 |    30/    0 batches | ms/batch 15.36 | loss  0.00017\n",
      "| epoch  12 |    40/    0 batches | ms/batch 14.83 | loss  0.00009\n",
      "| epoch  12 |    50/    0 batches | ms/batch 15.22 | loss  0.00012\n",
      "| epoch  12 |    60/    0 batches | ms/batch 15.05 | loss  0.00013\n",
      "| epoch  13 |    10/    0 batches | ms/batch 18.09 | loss  0.00011\n",
      "| epoch  13 |    20/    0 batches | ms/batch 15.10 | loss  0.00012\n",
      "| epoch  13 |    30/    0 batches | ms/batch 15.57 | loss  0.00011\n",
      "| epoch  13 |    40/    0 batches | ms/batch 18.18 | loss  0.00018\n",
      "| epoch  13 |    50/    0 batches | ms/batch 15.88 | loss  0.00011\n",
      "| epoch  13 |    60/    0 batches | ms/batch 15.80 | loss  0.00011\n",
      "| epoch  14 |    10/    0 batches | ms/batch 17.59 | loss  0.00012\n",
      "| epoch  14 |    20/    0 batches | ms/batch 15.61 | loss  0.00012\n",
      "| epoch  14 |    30/    0 batches | ms/batch 15.46 | loss  0.00011\n",
      "| epoch  14 |    40/    0 batches | ms/batch 14.86 | loss  0.00015\n",
      "| epoch  14 |    50/    0 batches | ms/batch 15.14 | loss  0.00013\n",
      "| epoch  14 |    60/    0 batches | ms/batch 15.19 | loss  0.00011\n",
      "| epoch  15 |    10/    0 batches | ms/batch 18.51 | loss  0.00014\n",
      "| epoch  15 |    20/    0 batches | ms/batch 16.24 | loss  0.00011\n",
      "| epoch  15 |    30/    0 batches | ms/batch 17.31 | loss  0.00011\n",
      "| epoch  15 |    40/    0 batches | ms/batch 16.73 | loss  0.00015\n",
      "| epoch  15 |    50/    0 batches | ms/batch 16.47 | loss  0.00012\n",
      "| epoch  15 |    60/    0 batches | ms/batch 16.19 | loss  0.00011\n",
      "| epoch  16 |    10/    0 batches | ms/batch 16.62 | loss  0.00012\n",
      "| epoch  16 |    20/    0 batches | ms/batch 14.99 | loss  0.00012\n",
      "| epoch  16 |    30/    0 batches | ms/batch 14.91 | loss  0.00010\n",
      "| epoch  16 |    40/    0 batches | ms/batch 15.10 | loss  0.00012\n",
      "| epoch  16 |    50/    0 batches | ms/batch 14.93 | loss  0.00011\n",
      "| epoch  16 |    60/    0 batches | ms/batch 15.08 | loss  0.00011\n",
      "| epoch  17 |    10/    0 batches | ms/batch 17.74 | loss  0.00011\n",
      "| epoch  17 |    20/    0 batches | ms/batch 15.04 | loss  0.00012\n",
      "| epoch  17 |    30/    0 batches | ms/batch 15.40 | loss  0.00011\n",
      "| epoch  17 |    40/    0 batches | ms/batch 15.32 | loss  0.00012\n",
      "| epoch  17 |    50/    0 batches | ms/batch 15.06 | loss  0.00012\n",
      "| epoch  17 |    60/    0 batches | ms/batch 15.07 | loss  0.00015\n",
      "| epoch  18 |    10/    0 batches | ms/batch 16.52 | loss  0.00011\n",
      "| epoch  18 |    20/    0 batches | ms/batch 15.14 | loss  0.00012\n",
      "| epoch  18 |    30/    0 batches | ms/batch 18.08 | loss  0.00010\n",
      "| epoch  18 |    40/    0 batches | ms/batch 15.96 | loss  0.00016\n",
      "| epoch  18 |    50/    0 batches | ms/batch 15.50 | loss  0.00011\n",
      "| epoch  18 |    60/    0 batches | ms/batch 15.80 | loss  0.00012\n",
      "| epoch  19 |    10/    0 batches | ms/batch 18.27 | loss  0.00013\n",
      "| epoch  19 |    20/    0 batches | ms/batch 16.33 | loss  0.00012\n",
      "| epoch  19 |    30/    0 batches | ms/batch 16.99 | loss  0.00015\n",
      "| epoch  19 |    40/    0 batches | ms/batch 16.18 | loss  0.00012\n",
      "| epoch  19 |    50/    0 batches | ms/batch 15.46 | loss  0.00010\n",
      "| epoch  19 |    60/    0 batches | ms/batch 16.69 | loss  0.00011\n",
      "| epoch  20 |    10/    0 batches | ms/batch 18.88 | loss  0.00018\n",
      "| epoch  20 |    20/    0 batches | ms/batch 15.76 | loss  0.00011\n",
      "| epoch  20 |    30/    0 batches | ms/batch 17.45 | loss  0.00012\n",
      "| epoch  20 |    40/    0 batches | ms/batch 15.94 | loss  0.00011\n",
      "| epoch  20 |    50/    0 batches | ms/batch 15.65 | loss  0.00010\n",
      "| epoch  20 |    60/    0 batches | ms/batch 15.12 | loss  0.00010\n"
     ]
    }
   ],
   "source": [
    "#drug's target gene data\n",
    "drugGeneDataset=DrugTargetDataset(drug_features)\n",
    "drugGeneDataset_loader = DataLoader(drugGeneDataset, batch_size=64, shuffle=True)\n",
    "#learn\n",
    "drugGeneCompressor=geneCompressing(drugGeneDataset_loader, num_gene_compressed_drug)\n",
    "#save\n",
    "drugGeneCompressor.eval()\n",
    "drugGeneCompressed=np.array([drugGeneCompressor.cpu()._encoder(torch.FloatTensor(drugGeneDataset[d])).data.numpy() for d in range(num_drugs)])\n",
    "torch.save(drugGeneCompressor.state_dict(), f'{DATA_PATH}/models/drugGeneCompressor.p')\n",
    "pickle.dump(drugGeneCompressed, open(f'{DATA_PATH}/models/drugGeneCompressed.p', 'wb'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [],
   "source": [
    "drugGeneCompressed=pickle.load(open(f'{DATA_PATH}/models/drugGeneCompressed.p', 'rb'))\n",
    "drugGeneCompressed=torch.FloatTensor(drugGeneCompressed)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "| epoch   1 |     1/    0 batches | ms/batch 155.43 | loss  2.29992\n",
      "| epoch   2 |     1/    0 batches | ms/batch 152.85 | loss  3.31703\n",
      "| epoch   3 |     1/    0 batches | ms/batch 153.76 | loss  2.99331\n",
      "| epoch   4 |     1/    0 batches | ms/batch 161.74 | loss  2.58272\n",
      "| epoch   5 |     1/    0 batches | ms/batch 155.35 | loss  2.24655\n",
      "| epoch   6 |     1/    0 batches | ms/batch 155.01 | loss  2.06670\n",
      "| epoch   7 |     1/    0 batches | ms/batch 154.29 | loss  2.00173\n",
      "| epoch   8 |     1/    0 batches | ms/batch 152.88 | loss  1.86386\n",
      "| epoch   9 |     1/    0 batches | ms/batch 153.28 | loss  1.79390\n",
      "| epoch  10 |     1/    0 batches | ms/batch 154.21 | loss  1.67008\n",
      "| epoch  11 |     1/    0 batches | ms/batch 155.46 | loss  1.64380\n",
      "| epoch  12 |     1/    0 batches | ms/batch 154.56 | loss  1.58257\n",
      "| epoch  13 |     1/    0 batches | ms/batch 153.29 | loss  1.62956\n",
      "| epoch  14 |     1/    0 batches | ms/batch 151.98 | loss  1.54405\n",
      "| epoch  15 |     1/    0 batches | ms/batch 153.49 | loss  1.54732\n",
      "| epoch  16 |     1/    0 batches | ms/batch 152.70 | loss  1.48662\n",
      "| epoch  17 |     1/    0 batches | ms/batch 155.41 | loss  1.47165\n",
      "| epoch  18 |     1/    0 batches | ms/batch 153.39 | loss  1.50418\n",
      "| epoch  19 |     1/    0 batches | ms/batch 152.45 | loss  1.48205\n",
      "| epoch  20 |     1/    0 batches | ms/batch 152.89 | loss  1.50635\n"
     ]
    }
   ],
   "source": [
    "\n",
    "#cell line's gene expression data\n",
    "cellGeneDataset=CellGeneDataset(cell_features)\n",
    "cellGeneDataset_loader = DataLoader(cellGeneDataset, batch_size=64, shuffle=True)\n",
    "#learn\n",
    "cellGeneCompressor=geneCompressing(cellGeneDataset_loader,num_gene_compressed_cell, noise_weight=0.01, log_interval=1 )\n",
    "#save\n",
    "cellGeneCompressor.eval()\n",
    "cellGeneCompressed=np.array([cellGeneCompressor.cpu()._encoder(torch.FloatTensor(cellGeneDataset[d])).data.numpy() for d in range(num_celllines)])\n",
    "torch.save(cellGeneCompressor.state_dict(), f'{DATA_PATH}/models/cellGeneCompressor.p')\n",
    "pickle.dump(cellGeneCompressed, open(f'{DATA_PATH}/models/cellGeneCompressed.p', 'wb'))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [],
   "source": [
    "cellGeneCompressed=pickle.load(open(f'{DATA_PATH}/models/cellGeneCompressed.p', 'rb'))\n",
    "cellGeneCompressed=torch.FloatTensor(cellGeneCompressed)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### <a id='toc1_4_3_4_'></a>[Synergy prediction](#toc0_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### <a id='toc1_4_3_4_1_'></a>[Train/test split in cross or external validation](#toc0_)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_cell_of_interest(tissues):\n",
    "    tissues_of_interests = [codes['tissue'].item2idx[minor_tissue] for minor_tissue in tissues]\n",
    "    cell_of_interest = cell_features.index[cell_features['tissue_id'].isin(tissues_of_interests)].tolist()\n",
    "    return cell_of_interest"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Train/test for general model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {},
   "outputs": [],
   "source": [
    "minor_tissues=['bone', 'prostate' ]\n",
    "cell_of_interest = get_cell_of_interest(minor_tissues)\n",
    "df_tissue_of_interest = df.loc[df['cell_line_name'].isin(cell_of_interest),:]\n",
    "df_all = df.drop(df_tissue_of_interest.index)\n",
    "#specific database\n",
    "#df_all=df_all.loc[df_all['study_id']==3]\n",
    "#cross validation\n",
    "df_train, df_test = train_test_split(df_all, test_size=0.2) #cross validation\n",
    "#external validation\n",
    "#df_train=df_all.loc[df_all['study_id']==3] 3: 'ALMANAC'\n",
    "#df_test=df_all.loc[df_all['study_id']==1] 1: 'ONEIL'"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Train/test for bone"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {},
   "outputs": [],
   "source": [
    "_df_bone = df.loc[df['cell_line_name'].isin(get_cell_of_interest(['bone'])),:]\n",
    "# Cross validation\n",
    "_df_train_bone, _df_test_bone = train_test_split(_df_bone, test_size=0.2, random_state=1)\n",
    "# External validation\n",
    "#_df_train_bone=_df_bone.loc[_df_bone['study_id']!=9]\n",
    "#_df_test_bone= _df_bone.loc[_df_bone['study_id']==9]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Train/test for prostate"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {},
   "outputs": [],
   "source": [
    "_df_prostate= df.loc[df['cell_line_name'].isin(get_cell_of_interest(['prostate'])),:]\n",
    "#Cross validation\n",
    "_df_train_prostate, _df_test_prostate = train_test_split(_df_prostate, test_size=0.2, random_state=1)\n",
    "# External validation\n",
    "#_df_train_prostate=_df_prostate.loc[_df_prostate['study_id']!=1]\n",
    "#_df_test_prostate=_df_prostate.loc[_df_prostate['study_id']==1]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### <a id='toc1_4_3_4_2_'></a>[Dataset and Dataloader](#toc0_)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {},
   "outputs": [],
   "source": [
    "class DrugCombDataset(Dataset):\n",
    "    def __init__(self, df, drug_features, cell_features):\n",
    "        self.df = df\n",
    "        self.drug_features = drug_features\n",
    "        self.cell_features = cell_features\n",
    "        \n",
    "    def __len__(self):\n",
    "        return len(self.df)\n",
    "    \n",
    "    def __getitem__(self, idx):\n",
    "        \n",
    "        d1 = self.df.iloc[idx, 0]\n",
    "        d2 = self.df.iloc[idx, 1]\n",
    "        cell = self.df.iloc[idx,2]\n",
    "        ri_d1 = 1.0 if self.df.iloc[idx,3] >ri_threshold else 0\n",
    "        ri_d2 = 1.0 if self.df.iloc[idx,4] >ri_threshold else 0\n",
    "        syn = 1.0 if self.df.iloc[idx, 5] >syn_threshold else 0\n",
    "        \n",
    "        \n",
    "        #external features\n",
    "        d1_fp = np.array(self.drug_features.loc[d1, 'fps'])\n",
    "        d1_sm = self.drug_features.loc[d1, 'smiles']\n",
    "        d1_sm = np.pad(d1_sm, pad_width=(0, max_drug_sm_len-len(d1_sm)), mode='constant', constant_values=0)\n",
    "        d1_gn=drugGeneCompressed[d1]\n",
    "        \n",
    "        d2_fp = np.array(self.drug_features.loc[d2, 'fps'])\n",
    "        d2_sm = self.drug_features.loc[d2, 'smiles']\n",
    "        d2_sm = np.pad(d2_sm, pad_width=(0, max_drug_sm_len-len(d2_sm)), mode='constant', constant_values=0)\n",
    "        d2_gn=drugGeneCompressed[d2]\n",
    "        \n",
    "        c_ts = self.cell_features.loc[cell, 'tissue_id']\n",
    "        c_ds = self.cell_features.loc[cell, 'disease_id']\n",
    "        c_gn= cellGeneCompressed[cell]\n",
    "        \n",
    "        sample = {\n",
    "            'd1': d1,\n",
    "            'd1_fp': d1_fp,\n",
    "            'd1_sm': d1_sm,\n",
    "            'd1_gn': d1_gn,\n",
    "            \n",
    "            'd2': d2,\n",
    "            'd2_fp': d2_fp,\n",
    "            'd2_sm': d2_sm,\n",
    "            'd2_gn': d2_gn,\n",
    "            \n",
    "            'cell': cell,\n",
    "            'c_ts': c_ts,\n",
    "            'c_ds': c_ds, #missing -1\n",
    "            'c_gn': c_gn,\n",
    "            \n",
    "            'ri_d1': ri_d1,\n",
    "            'ri_d2': ri_d2,\n",
    "            'syn': syn\n",
    "        }\n",
    "        \n",
    "        return sample"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### <a id='toc1_4_3_4_3_'></a>[Load dataset](#toc0_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "General model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {},
   "outputs": [],
   "source": [
    "train = DrugCombDataset(df_train, drug_features, cell_features)\n",
    "train_loader = DataLoader(train, batch_size=bsz, shuffle=True )\n",
    "test = DrugCombDataset(df_test, drug_features, cell_features)\n",
    "test_loader = DataLoader(test, batch_size=bsz, shuffle=True )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Bone"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {},
   "outputs": [],
   "source": [
    "_train_bone = DrugCombDataset(_df_train_bone, drug_features, cell_features)\n",
    "_train_loader_bone = DataLoader(_train_bone, batch_size=bsz, shuffle=True )\n",
    "_test_bone = DrugCombDataset(_df_test_bone, drug_features, cell_features)\n",
    "_test_loader_bone = DataLoader(_test_bone, batch_size=bsz, shuffle=False )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Prostate"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "metadata": {},
   "outputs": [],
   "source": [
    "_train_prostate = DrugCombDataset(_df_train_prostate, drug_features, cell_features)\n",
    "_train_loader_prostate = DataLoader(_train_prostate, batch_size=bsz, shuffle=True )\n",
    "_test_prostate = DrugCombDataset(_df_test_prostate, drug_features, cell_features)\n",
    "_test_loader_prostate = DataLoader(_test_prostate, batch_size=bsz, shuffle=False )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### <a id='toc1_4_3_5_'></a>[Prediction model](#toc0_)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "metadata": {},
   "outputs": [],
   "source": [
    "class DrugEncoder(nn.Module):\n",
    "    def __init__(self,\n",
    "                 num_drugs=num_drugs,\n",
    "                 num_ID_emb=0,\n",
    "                 num_drug_fp=num_drug_fp,\n",
    "                 max_drug_sm_len=max_drug_sm_len,\n",
    "                 num_gene = num_gene_compressed_drug,\n",
    "                 num_comp_char=len(codes['mole'].idx2item),\n",
    "                 fp_embed_sz = 32,\n",
    "                 gene_embed_sz = int(num_gene_compressed_drug/2),\n",
    "                 out_size=64,\n",
    "                 dropout=0.3):\n",
    "        super(DrugEncoder, self).__init__()\n",
    "        \n",
    "        self.dropout= dropout\n",
    "        #DRUG\n",
    "        #drug ID\n",
    "        #self.embed_id = nn.Embedding(num_drugs, num_ID_emb)\n",
    "        \n",
    "        #compound ID\n",
    "        self.embed_comp = nn.Embedding(num_comp_char, num_comp_char, padding_idx=0)#padding's idx=0\n",
    "        #encoding compound\n",
    "        self.encoderlayer = nn.TransformerEncoderLayer(d_model=num_comp_char, nhead=4)\n",
    "        self.encoder = nn.TransformerEncoder(self.encoderlayer, num_layers=1)\n",
    "        \n",
    "        #fingerprint\n",
    "        self.dense_fp = nn.Linear(num_drug_fp,fp_embed_sz)\n",
    "        #gene\n",
    "        self.dense_gene = nn.Linear(num_gene,gene_embed_sz)\n",
    "        \n",
    "        #depthwise for compound encoding\n",
    "        self.conv = nn.Conv2d(1, 1, (1, num_comp_char), groups=1)\n",
    "        \n",
    "        #combined\n",
    "        combined_sz = num_ID_emb+fp_embed_sz+max_drug_sm_len+gene_embed_sz\n",
    "        self.FC2 = FC2(combined_sz, out_size, dropout)\n",
    "\n",
    "    def forward(self, d_list):\n",
    "        \"\"\"\n",
    "            id: bsz*1\n",
    "            fp: bsz*num_drug_fp\n",
    "            sm: bsz*max_drug_sm_len\n",
    "        \"\"\"\n",
    "        id, fp, sm, gn = d_list\n",
    "        \n",
    "        sm = self.embed_comp(sm) #bsz*max_drug_sm_len*num_comp_char(embedding size)\n",
    "        sm = self.encoder(sm)\n",
    "        sm = self.conv(sm.unsqueeze(1)).squeeze()\n",
    "        \n",
    "        fp = F.relu(self.dense_fp(fp))\n",
    "        gn = F.relu(self.dense_gene(gn))\n",
    "        \n",
    "        #combine\n",
    "        x = torch.cat((fp, sm, gn),1) # bsz*[num_emb_id+num_drug_fp+max+drug_sm]\n",
    "        x = self.FC2(x)\n",
    "        \n",
    "        return x\n",
    "        "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "metadata": {},
   "outputs": [],
   "source": [
    "class CellEncoder(nn.Module):\n",
    "    def __init__(self,\n",
    "                 num_cells=num_celllines,\n",
    "                 num_tissue=0,\n",
    "                 num_disease=num_disease,\n",
    "                 num_ID_emb=0,\n",
    "                 gene_embed_sz=int(num_gene_compressed_cell/2),\n",
    "                 num_gene=num_gene_compressed_cell,\n",
    "                 out_size=64,\n",
    "                 dropout=0.3):\n",
    "        super(CellEncoder, self).__init__()\n",
    "        \n",
    "        self.dropout= dropout\n",
    "        #cell ID\n",
    "        #self.embed_id = nn.Embedding(num_cells, num_ID_emb)\n",
    "        #cell tissue\n",
    "        #self.embed_ts = nn.Embedding(num_tissue, num_tissue)\n",
    "        #cell disease\n",
    "        self.embed_ds = nn.Embedding(num_disease, num_disease, padding_idx=3)\n",
    "        #gene\n",
    "        self.dense_gene = nn.Linear(num_gene,gene_embed_sz)\n",
    "        \n",
    "        #combined\n",
    "        combined_sz = num_ID_emb+num_tissue+num_disease+gene_embed_sz\n",
    "        self.FC2 = FC2(combined_sz, out_size, dropout)\n",
    "    \n",
    "        \n",
    "    def forward(self, c_list):\n",
    "        \"\"\"\n",
    "            id: bsz*1\n",
    "            fp: bsz*num_drug_fp\n",
    "            sm: bsz*max_drug_sm_len\n",
    "        \"\"\"\n",
    "        id, ts, ds, gn = c_list\n",
    "        ds = F.relu(self.embed_ds(ds)) #bsz*num_diesaes\n",
    "        \n",
    "        gn = F.relu(self.dense_gene(gn)) #bsz*gene_embed_sz\n",
    "        \n",
    "        #combine\n",
    "        x = torch.cat((ds, gn),1) # bsz*combined_sz\n",
    "        x = self.FC2(x)\n",
    "\n",
    "        return x\n",
    "        "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Comb(nn.Module):\n",
    "    def __init__(self, num_cells=num_celllines, \n",
    "                 num_drugs=num_drugs,\n",
    "                 num_drug_fp=num_drug_fp,\n",
    "                 max_drug_sm_len=max_drug_sm_len,\n",
    "                num_comp_char=len(codes['mole'].idx2item),\n",
    "                 num_ID_emb=0,\n",
    "                 out_size=64,\n",
    "                dropout=0.3):\n",
    "        \n",
    "        super(Comb, self).__init__()\n",
    "        \n",
    "        self.dropout=dropout    \n",
    "        #drug\n",
    "        self.drugEncoder = DrugEncoder()\n",
    "        #cell\n",
    "        self.cellEncoder = CellEncoder()\n",
    "        #fc\n",
    "        self.fc_syn = FC2(out_size*3, 1, dropout)\n",
    "        self.fc_ri = FC2(out_size*2, 1, dropout)\n",
    "        \n",
    "    def forward(self, d1_list, d2_list, c_list):\n",
    "        d1 = self.drugEncoder(d1_list)\n",
    "        d2 = self.drugEncoder(d2_list)\n",
    "        c = self.cellEncoder(c_list)\n",
    "        \n",
    "        syn = self.fc_syn(torch.cat((d1, d2, c),1))\n",
    "        ri1 = self.fc_ri(torch.cat((d1,c),1))\n",
    "        ri2 = self.fc_ri(torch.cat((d2,c),1))\n",
    "        \n",
    "        return syn, ri1, ri2\n",
    "    "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### <a id='toc1_4_3_6_'></a>[Training](#toc0_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The model employs a multi-task learning framework, meaning it simultaneously learns to predict both monotherapy sensitivity and drug combination synergy. This approach helps the model leverage commonalities between the two tasks to improve its predictive accuracy. The training loss is a combination of mean squared error (MSE) for the regression tasks (predicting the exact scores of sensitivity and synergy) and binary cross-entropy for classification tasks (predicting whether a drug combination is above a certain synergy threshold).\n",
    "\n",
    "Transfer learning is a crucial aspect of the training process. The model first learns from data-rich tissues and then transfers its learned parameters to models targeting understudied tissues. This transfer is fine-tuned with data from the understudied tissues to adapt the model to their specific characteristics."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div align=\"center\">\n",
    "    <img src=\"public/assets/transfer_learning_figure.png\" width=\"700\">\n",
    "    <p><b>Figure 4:</b> Transfer Learning Diagram from <a href=\"https://doi.org/10.1101/2020.02.05.932657\">Kim et al. (2020)</a></p>\n",
    "</div>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### <a id='toc1_4_3_6_1_'></a>[Hyperparameters](#toc0_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We defined the hyperparameters above in Cell 58. These hyperparameters are the same as those used by [Kim et al. (2020)](https://doi.org/10.1101/2020.02.05.932657). This includes 10 total epochs for the general model, a batch size of 128, and a dropout of 0.3. For more information on the default hyperparameters, you can find them all in Cell 58 above in the [Configuration](#configuration) section."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### <a id='toc1_4_3_6_2_'></a>[Computational Requirements](#toc0_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In terms of the computational feasibility of this paper, you should be able to perform all data preprocessing and training on your local machine, without a need for any additional computing clusters, assuming you have a GPU. In my own testing, I trained the model for 20 epochs with two different GPUs to compare performance on high versus medium-end consumer GPUs. Specifically, I first ran this notebook using an NVIDIA GTX 4090 GPU, which had no issues at all during training. The average minutes per epoch (m/e) using this hardware was ~3 minutes, for a total runtime of ~30 minutes. I also tested this on a NVIDIA GTX 2060 SUPER GPU, which obviously performed worse in comparison, but not prohibitively so. The 2060 SUPER completed training with 5.5 m/e and a total runtime of ~55 minutes. All tests were carried out using the same conda environment and on Ubuntu 22.04."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/blake/miniconda3/envs/synergy/lib/python3.8/site-packages/torch/nn/modules/transformer.py:286: UserWarning: enable_nested_tensor is True, but self.use_nested_tensor is False because encoder_layer.self_attn.batch_first was not True(use batch_first for better inference performance)\n",
      "  warnings.warn(f\"enable_nested_tensor is True, but self.use_nested_tensor is False because {why_not_sparsity_fast_path}\")\n"
     ]
    }
   ],
   "source": [
    "model = Comb()\n",
    "if cuda:\n",
    "    model = model.cuda()\n",
    "#Regression\n",
    "#criterion_mse = nn.MSELoss()\n",
    "#Classification\n",
    "criterion_bce = nn.BCEWithLogitsLoss()\n",
    "optimizer = optim.Adagrad(model.parameters())\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Training\n",
    "def training(isAux, data_loader):\n",
    "    model.train()\n",
    "    total_loss = 0\n",
    "    start_time = time.time()\n",
    "\n",
    "    for iteration, sample in enumerate(data_loader):\n",
    "        d1=Variable(sample['d1'])\n",
    "        d1_fp = Variable(sample['d1_fp'].float())\n",
    "        d1_sm = Variable(sample['d1_sm'])\n",
    "        d1_gn = Variable(sample['d1_gn'].float())\n",
    "        \n",
    "        d2=Variable(sample['d2'])\n",
    "        d2_fp = Variable(sample['d2_fp'].float())\n",
    "        d2_sm = Variable(sample['d2_sm'])\n",
    "        d2_gn = Variable(sample['d2_gn'].float())\n",
    "        \n",
    "        cell = Variable(sample['cell'])\n",
    "        c_ts = Variable(sample['c_ts'])\n",
    "        c_ds = Variable(sample['c_ds'])\n",
    "        c_gn = Variable(sample['c_gn'].float())\n",
    "        \n",
    "        syn_true = Variable(sample['syn'].float())\n",
    "        ri_d1=Variable(sample['ri_d1'].float())\n",
    "        ri_d2=Variable(sample['ri_d2'].float())\n",
    "\n",
    "\n",
    "        if cuda:\n",
    "            d1=d1.cuda()\n",
    "            d1_fp=d1_fp.cuda()\n",
    "            d1_sm=d1_sm.cuda()\n",
    "            d1_gn=d1_gn.cuda()\n",
    "            \n",
    "            d2=d2.cuda()\n",
    "            d2_fp=d2_fp.cuda()\n",
    "            d2_sm=d2_sm.cuda()\n",
    "            d2_gn=d2_gn.cuda()\n",
    "            \n",
    "            cell=cell.cuda()\n",
    "            c_ts=c_ts.cuda()\n",
    "            c_ds=c_ds.cuda()\n",
    "            c_gn=c_gn.cuda()\n",
    "            \n",
    "            syn_true=syn_true.cuda()\n",
    "            ri_d1=ri_d1.cuda()\n",
    "            ri_d2=ri_d2.cuda()\n",
    "\n",
    "        optimizer.zero_grad()\n",
    "        \n",
    "        syn, ri1, ri2 = model((d1, d1_fp, d1_sm, d1_gn), (d2, d2_fp, d2_sm, d2_gn), (cell, c_ts, c_ds, c_gn) )\n",
    "        \n",
    "        if not isAux:\n",
    "            loss = criterion_bce(syn, syn_true.view(-1,1))\n",
    "        else:\n",
    "            loss = criterion_bce(ri1, ri_d1.view(-1,1))+criterion_bce(ri2, ri_d2.view(-1,1))\n",
    "            \n",
    "        loss.backward()\n",
    "        optimizer.step()\n",
    "        total_loss += loss.data\n",
    "\n",
    "        if iteration % log_interval == 0 and iteration > 0:\n",
    "            cur_loss = total_loss.item() / log_interval\n",
    "            elapsed = time.time() - start_time\n",
    "            print('| epoch {:3d} | {:5d}/{:5d} batches | ms/batch {:5.2f} | loss {:8.5f}'.format(epoch, iteration, int(len(train_loader)/bsz), elapsed * 1000/log_interval, cur_loss))\n",
    "\n",
    "            total_loss = 0\n",
    "            start_time = time.time()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "metadata": {},
   "outputs": [],
   "source": [
    "def evaluate(data_loader):\n",
    "    model.eval()\n",
    "    total_loss = 0\n",
    "    total_loss_sen = 0\n",
    "\n",
    "    \n",
    "    #loss\n",
    "    with torch.no_grad():\n",
    "        for iteration, sample in enumerate(data_loader):\n",
    "            d1=Variable(sample['d1'])\n",
    "            d1_fp = Variable(sample['d1_fp'].float())\n",
    "            d1_sm = Variable(sample['d1_sm'])\n",
    "            d1_gn = Variable(sample['d1_gn'].float())\n",
    "\n",
    "            d2=Variable(sample['d2'])\n",
    "            d2_fp = Variable(sample['d2_fp'].float())\n",
    "            d2_sm = Variable(sample['d2_sm'])\n",
    "            d2_gn = Variable(sample['d2_gn'].float())\n",
    "\n",
    "            cell = Variable(sample['cell'])\n",
    "            c_ts = Variable(sample['c_ts'])\n",
    "            c_ds = Variable(sample['c_ds'])\n",
    "            c_gn = Variable(sample['c_gn'].float())\n",
    "\n",
    "            syn_true = Variable(sample['syn'].float())\n",
    "            ri_d1=Variable(sample['ri_d1'].float())\n",
    "            ri_d2=Variable(sample['ri_d2'].float())\n",
    "\n",
    "\n",
    "            if cuda:\n",
    "                d1=d1.cuda()\n",
    "                d1_fp=d1_fp.cuda()\n",
    "                d1_sm=d1_sm.cuda()\n",
    "                d1_gn=d1_gn.cuda()\n",
    "\n",
    "                d2=d2.cuda()\n",
    "                d2_fp=d2_fp.cuda()\n",
    "                d2_sm=d2_sm.cuda()\n",
    "                d2_gn=d2_gn.cuda()\n",
    "\n",
    "                cell=cell.cuda()\n",
    "                c_ts=c_ts.cuda()\n",
    "                c_ds=c_ds.cuda()\n",
    "                c_gn=c_gn.cuda()\n",
    "\n",
    "                syn_true=syn_true.cuda()\n",
    "                ri_d1=ri_d1.cuda()\n",
    "                ri_d2=ri_d2.cuda()\n",
    "\n",
    "\n",
    "            syn,ri1,ri2 = model((d1, d1_fp, d1_sm, d1_gn), (d2, d2_fp, d2_sm, d2_gn), (cell, c_ts,c_ds,c_gn) )\n",
    "            loss = criterion_bce(syn, syn_true.view(-1,1))\n",
    "            total_loss +=loss.data\n",
    "            loss_sen = (criterion_bce(ri1, ri_d1.view(-1,1))+criterion_bce(ri2, ri_d2.view(-1,1)))/2\n",
    "            total_loss_sen += loss_sen.data\n",
    "\n",
    "        print('syn mse', total_loss.item()/(iteration+1))\n",
    "        print('sen_mse', total_loss_sen.item()/(iteration+1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "| epoch   1 |   100/   18 batches | ms/batch 62.58 | loss  0.38068\n",
      "| epoch   1 |   200/   18 batches | ms/batch 61.15 | loss  0.33605\n",
      "| epoch   1 |   300/   18 batches | ms/batch 59.03 | loss  0.31480\n",
      "| epoch   1 |   400/   18 batches | ms/batch 61.55 | loss  0.30493\n",
      "| epoch   1 |   500/   18 batches | ms/batch 58.84 | loss  0.29916\n",
      "| epoch   1 |   600/   18 batches | ms/batch 58.40 | loss  0.28708\n",
      "| epoch   1 |   700/   18 batches | ms/batch 58.45 | loss  0.28399\n",
      "| epoch   1 |   800/   18 batches | ms/batch 61.39 | loss  0.28293\n",
      "| epoch   1 |   900/   18 batches | ms/batch 58.60 | loss  0.27937\n",
      "| epoch   1 |  1000/   18 batches | ms/batch 58.79 | loss  0.28504\n",
      "| epoch   1 |  1100/   18 batches | ms/batch 58.81 | loss  0.27313\n",
      "| epoch   1 |  1200/   18 batches | ms/batch 62.14 | loss  0.27800\n",
      "| epoch   1 |  1300/   18 batches | ms/batch 59.00 | loss  0.27238\n",
      "| epoch   1 |  1400/   18 batches | ms/batch 59.12 | loss  0.25997\n",
      "| epoch   1 |  1500/   18 batches | ms/batch 59.29 | loss  0.26617\n",
      "| epoch   1 |  1600/   18 batches | ms/batch 62.92 | loss  0.25985\n",
      "| epoch   1 |  1700/   18 batches | ms/batch 59.36 | loss  0.26113\n",
      "| epoch   1 |  1800/   18 batches | ms/batch 59.50 | loss  0.26331\n",
      "| epoch   1 |  1900/   18 batches | ms/batch 59.59 | loss  0.25985\n",
      "| epoch   1 |  2000/   18 batches | ms/batch 60.32 | loss  0.25813\n",
      "| epoch   1 |  2100/   18 batches | ms/batch 60.43 | loss  0.26029\n",
      "| epoch   1 |  2200/   18 batches | ms/batch 64.81 | loss  0.25321\n",
      "| epoch   1 |  2300/   18 batches | ms/batch 61.41 | loss  0.25466\n",
      "| epoch   1 |   100/   18 batches | ms/batch 62.48 | loss  0.27817\n",
      "| epoch   1 |   200/   18 batches | ms/batch 61.90 | loss  0.19894\n",
      "| epoch   1 |   300/   18 batches | ms/batch 62.01 | loss  0.20873\n",
      "| epoch   1 |   400/   18 batches | ms/batch 62.08 | loss  0.18356\n",
      "| epoch   1 |   500/   18 batches | ms/batch 62.27 | loss  0.19218\n",
      "| epoch   1 |   600/   18 batches | ms/batch 67.26 | loss  0.18990\n",
      "| epoch   1 |   700/   18 batches | ms/batch 62.50 | loss  0.18233\n",
      "| epoch   1 |   800/   18 batches | ms/batch 62.83 | loss  0.19936\n",
      "| epoch   1 |   900/   18 batches | ms/batch 62.95 | loss  0.19113\n",
      "| epoch   1 |  1000/   18 batches | ms/batch 63.43 | loss  0.18770\n",
      "| epoch   1 |  1100/   18 batches | ms/batch 63.49 | loss  0.19279\n",
      "| epoch   1 |  1200/   18 batches | ms/batch 63.34 | loss  0.19413\n",
      "| epoch   1 |  1300/   18 batches | ms/batch 63.45 | loss  0.19312\n",
      "| epoch   1 |  1400/   18 batches | ms/batch 63.30 | loss  0.19516\n",
      "| epoch   1 |  1500/   18 batches | ms/batch 68.85 | loss  0.18924\n",
      "| epoch   1 |  1600/   18 batches | ms/batch 63.43 | loss  0.19128\n",
      "| epoch   1 |  1700/   18 batches | ms/batch 63.40 | loss  0.18930\n",
      "| epoch   1 |  1800/   18 batches | ms/batch 63.26 | loss  0.18812\n",
      "| epoch   1 |  1900/   18 batches | ms/batch 63.34 | loss  0.18743\n",
      "| epoch   1 |  2000/   18 batches | ms/batch 63.19 | loss  0.18539\n",
      "| epoch   1 |  2100/   18 batches | ms/batch 63.24 | loss  0.18334\n",
      "| epoch   1 |  2200/   18 batches | ms/batch 63.34 | loss  0.18253\n",
      "| epoch   1 |  2300/   18 batches | ms/batch 63.24 | loss  0.18551\n",
      "syn mse 0.2542644559021514\n",
      "sen_mse 0.0890474173735277\n",
      "-----------------------------------------------------------------------------------------\n",
      "| epoch   2 |   100/   18 batches | ms/batch 63.97 | loss  0.27243\n",
      "| epoch   2 |   200/   18 batches | ms/batch 63.21 | loss  0.26230\n",
      "| epoch   2 |   300/   18 batches | ms/batch 63.29 | loss  0.26228\n",
      "| epoch   2 |   400/   18 batches | ms/batch 63.18 | loss  0.24879\n",
      "| epoch   2 |   500/   18 batches | ms/batch 63.22 | loss  0.24909\n",
      "| epoch   2 |   600/   18 batches | ms/batch 63.25 | loss  0.25628\n",
      "| epoch   2 |   700/   18 batches | ms/batch 63.45 | loss  0.25072\n",
      "| epoch   2 |   800/   18 batches | ms/batch 63.56 | loss  0.25410\n",
      "| epoch   2 |   900/   18 batches | ms/batch 63.15 | loss  0.24968\n",
      "| epoch   2 |  1000/   18 batches | ms/batch 70.29 | loss  0.24893\n",
      "| epoch   2 |  1100/   18 batches | ms/batch 62.92 | loss  0.25495\n",
      "| epoch   2 |  1200/   18 batches | ms/batch 63.00 | loss  0.25666\n",
      "| epoch   2 |  1300/   18 batches | ms/batch 63.04 | loss  0.25567\n",
      "| epoch   2 |  1400/   18 batches | ms/batch 63.07 | loss  0.24263\n",
      "| epoch   2 |  1500/   18 batches | ms/batch 63.20 | loss  0.24543\n",
      "| epoch   2 |  1600/   18 batches | ms/batch 63.28 | loss  0.24239\n",
      "| epoch   2 |  1700/   18 batches | ms/batch 63.25 | loss  0.23940\n",
      "| epoch   2 |  1800/   18 batches | ms/batch 63.70 | loss  0.23778\n",
      "| epoch   2 |  1900/   18 batches | ms/batch 64.20 | loss  0.24348\n",
      "| epoch   2 |  2000/   18 batches | ms/batch 64.20 | loss  0.24137\n",
      "| epoch   2 |  2100/   18 batches | ms/batch 64.13 | loss  0.24327\n",
      "| epoch   2 |  2200/   18 batches | ms/batch 63.77 | loss  0.24162\n",
      "| epoch   2 |  2300/   18 batches | ms/batch 63.42 | loss  0.24502\n",
      "| epoch   2 |   100/   18 batches | ms/batch 64.38 | loss  0.18045\n",
      "| epoch   2 |   200/   18 batches | ms/batch 63.64 | loss  0.19188\n",
      "| epoch   2 |   300/   18 batches | ms/batch 72.48 | loss  0.18953\n",
      "| epoch   2 |   400/   18 batches | ms/batch 63.72 | loss  0.18995\n",
      "| epoch   2 |   500/   18 batches | ms/batch 63.85 | loss  0.18735\n",
      "| epoch   2 |   600/   18 batches | ms/batch 63.99 | loss  0.18337\n",
      "| epoch   2 |   700/   18 batches | ms/batch 64.07 | loss  0.19097\n",
      "| epoch   2 |   800/   18 batches | ms/batch 64.03 | loss  0.18006\n",
      "| epoch   2 |   900/   18 batches | ms/batch 64.03 | loss  0.18266\n",
      "| epoch   2 |  1000/   18 batches | ms/batch 64.15 | loss  0.18987\n",
      "| epoch   2 |  1100/   18 batches | ms/batch 64.39 | loss  0.18352\n",
      "| epoch   2 |  1200/   18 batches | ms/batch 65.21 | loss  0.17555\n",
      "| epoch   2 |  1300/   18 batches | ms/batch 65.13 | loss  0.17115\n",
      "| epoch   2 |  1400/   18 batches | ms/batch 65.21 | loss  0.18102\n",
      "| epoch   2 |  1500/   18 batches | ms/batch 64.50 | loss  0.18782\n",
      "| epoch   2 |  1600/   18 batches | ms/batch 64.52 | loss  0.17842\n",
      "| epoch   2 |  1700/   18 batches | ms/batch 64.70 | loss  0.18034\n",
      "| epoch   2 |  1800/   18 batches | ms/batch 64.64 | loss  0.18799\n",
      "| epoch   2 |  1900/   18 batches | ms/batch 64.73 | loss  0.18178\n",
      "| epoch   2 |  2000/   18 batches | ms/batch 64.68 | loss  0.19336\n",
      "| epoch   2 |  2100/   18 batches | ms/batch 64.79 | loss  0.18208\n",
      "| epoch   2 |  2200/   18 batches | ms/batch 75.18 | loss  0.17151\n",
      "| epoch   2 |  2300/   18 batches | ms/batch 64.57 | loss  0.17630\n",
      "syn mse 0.22917623471323217\n",
      "sen_mse 0.08896537871433641\n",
      "-----------------------------------------------------------------------------------------\n",
      "| epoch   3 |   100/   18 batches | ms/batch 64.66 | loss  0.24733\n",
      "| epoch   3 |   200/   18 batches | ms/batch 64.34 | loss  0.24291\n",
      "| epoch   3 |   300/   18 batches | ms/batch 64.50 | loss  0.24256\n",
      "| epoch   3 |   400/   18 batches | ms/batch 64.71 | loss  0.23931\n",
      "| epoch   3 |   500/   18 batches | ms/batch 64.93 | loss  0.24627\n",
      "| epoch   3 |   600/   18 batches | ms/batch 64.88 | loss  0.23975\n",
      "| epoch   3 |   700/   18 batches | ms/batch 64.94 | loss  0.24486\n",
      "| epoch   3 |   800/   18 batches | ms/batch 65.05 | loss  0.23056\n",
      "| epoch   3 |   900/   18 batches | ms/batch 65.08 | loss  0.23620\n",
      "| epoch   3 |  1000/   18 batches | ms/batch 65.18 | loss  0.23782\n",
      "| epoch   3 |  1100/   18 batches | ms/batch 65.13 | loss  0.23638\n",
      "| epoch   3 |  1200/   18 batches | ms/batch 65.17 | loss  0.23869\n",
      "| epoch   3 |  1300/   18 batches | ms/batch 65.37 | loss  0.23074\n",
      "| epoch   3 |  1400/   18 batches | ms/batch 65.32 | loss  0.24049\n",
      "| epoch   3 |  1500/   18 batches | ms/batch 65.34 | loss  0.23918\n",
      "| epoch   3 |  1600/   18 batches | ms/batch 65.31 | loss  0.23694\n",
      "| epoch   3 |  1700/   18 batches | ms/batch 65.28 | loss  0.22922\n",
      "| epoch   3 |  1800/   18 batches | ms/batch 77.25 | loss  0.23295\n",
      "| epoch   3 |  1900/   18 batches | ms/batch 65.15 | loss  0.23485\n",
      "| epoch   3 |  2000/   18 batches | ms/batch 65.24 | loss  0.23448\n",
      "| epoch   3 |  2100/   18 batches | ms/batch 65.16 | loss  0.23309\n",
      "| epoch   3 |  2200/   18 batches | ms/batch 65.21 | loss  0.23687\n",
      "| epoch   3 |  2300/   18 batches | ms/batch 64.71 | loss  0.24009\n",
      "| epoch   3 |   100/   18 batches | ms/batch 64.78 | loss  0.18772\n",
      "| epoch   3 |   200/   18 batches | ms/batch 64.00 | loss  0.18010\n",
      "| epoch   3 |   300/   18 batches | ms/batch 64.02 | loss  0.18561\n",
      "| epoch   3 |   400/   18 batches | ms/batch 64.02 | loss  0.18719\n",
      "| epoch   3 |   500/   18 batches | ms/batch 64.10 | loss  0.17789\n",
      "| epoch   3 |   600/   18 batches | ms/batch 64.17 | loss  0.18376\n",
      "| epoch   3 |   700/   18 batches | ms/batch 64.27 | loss  0.18466\n",
      "| epoch   3 |   800/   18 batches | ms/batch 64.44 | loss  0.18449\n",
      "| epoch   3 |   900/   18 batches | ms/batch 64.91 | loss  0.18031\n",
      "| epoch   3 |  1000/   18 batches | ms/batch 64.95 | loss  0.18699\n",
      "| epoch   3 |  1100/   18 batches | ms/batch 64.88 | loss  0.17422\n",
      "| epoch   3 |  1200/   18 batches | ms/batch 65.06 | loss  0.18485\n",
      "| epoch   3 |  1300/   18 batches | ms/batch 64.54 | loss  0.17909\n",
      "| epoch   3 |  1400/   18 batches | ms/batch 64.60 | loss  0.18431\n",
      "| epoch   3 |  1500/   18 batches | ms/batch 64.49 | loss  0.17228\n",
      "| epoch   3 |  1600/   18 batches | ms/batch 64.45 | loss  0.18337\n",
      "| epoch   3 |  1700/   18 batches | ms/batch 64.49 | loss  0.19001\n",
      "| epoch   3 |  1800/   18 batches | ms/batch 64.50 | loss  0.17347\n",
      "| epoch   3 |  1900/   18 batches | ms/batch 64.57 | loss  0.16922\n",
      "| epoch   3 |  2000/   18 batches | ms/batch 64.56 | loss  0.16636\n",
      "| epoch   3 |  2100/   18 batches | ms/batch 64.51 | loss  0.17390\n",
      "| epoch   3 |  2200/   18 batches | ms/batch 64.52 | loss  0.17072\n",
      "| epoch   3 |  2300/   18 batches | ms/batch 78.79 | loss  0.17015\n",
      "syn mse 0.22457002338610196\n",
      "sen_mse 0.08815972630965487\n",
      "-----------------------------------------------------------------------------------------\n",
      "| epoch   4 |   100/   18 batches | ms/batch 65.00 | loss  0.24840\n",
      "| epoch   4 |   200/   18 batches | ms/batch 64.55 | loss  0.23020\n",
      "| epoch   4 |   300/   18 batches | ms/batch 64.76 | loss  0.23184\n",
      "| epoch   4 |   400/   18 batches | ms/batch 64.88 | loss  0.22434\n",
      "| epoch   4 |   500/   18 batches | ms/batch 64.93 | loss  0.23659\n",
      "| epoch   4 |   600/   18 batches | ms/batch 64.90 | loss  0.23267\n",
      "| epoch   4 |   700/   18 batches | ms/batch 65.39 | loss  0.23034\n",
      "| epoch   4 |   800/   18 batches | ms/batch 65.88 | loss  0.23289\n",
      "| epoch   4 |   900/   18 batches | ms/batch 65.98 | loss  0.22784\n",
      "| epoch   4 |  1000/   18 batches | ms/batch 66.00 | loss  0.23778\n",
      "| epoch   4 |  1100/   18 batches | ms/batch 66.09 | loss  0.23858\n",
      "| epoch   4 |  1200/   18 batches | ms/batch 65.98 | loss  0.23268\n",
      "| epoch   4 |  1300/   18 batches | ms/batch 65.96 | loss  0.22390\n",
      "| epoch   4 |  1400/   18 batches | ms/batch 65.24 | loss  0.23180\n",
      "| epoch   4 |  1500/   18 batches | ms/batch 64.83 | loss  0.23081\n",
      "| epoch   4 |  1600/   18 batches | ms/batch 65.11 | loss  0.23661\n",
      "| epoch   4 |  1700/   18 batches | ms/batch 64.88 | loss  0.23334\n",
      "| epoch   4 |  1800/   18 batches | ms/batch 64.99 | loss  0.22580\n",
      "| epoch   4 |  1900/   18 batches | ms/batch 64.84 | loss  0.22550\n",
      "| epoch   4 |  2000/   18 batches | ms/batch 64.82 | loss  0.23151\n",
      "| epoch   4 |  2100/   18 batches | ms/batch 64.87 | loss  0.22961\n",
      "| epoch   4 |  2200/   18 batches | ms/batch 64.95 | loss  0.23122\n",
      "| epoch   4 |  2300/   18 batches | ms/batch 64.87 | loss  0.22916\n",
      "| epoch   4 |   100/   18 batches | ms/batch 65.74 | loss  0.19166\n",
      "| epoch   4 |   200/   18 batches | ms/batch 64.89 | loss  0.17918\n",
      "| epoch   4 |   300/   18 batches | ms/batch 64.89 | loss  0.18517\n",
      "| epoch   4 |   400/   18 batches | ms/batch 64.81 | loss  0.16604\n",
      "| epoch   4 |   500/   18 batches | ms/batch 64.92 | loss  0.17947\n",
      "| epoch   4 |   600/   18 batches | ms/batch 82.24 | loss  0.16836\n",
      "| epoch   4 |   700/   18 batches | ms/batch 64.62 | loss  0.18693\n",
      "| epoch   4 |   800/   18 batches | ms/batch 64.67 | loss  0.18315\n",
      "| epoch   4 |   900/   18 batches | ms/batch 64.68 | loss  0.17209\n",
      "| epoch   4 |  1000/   18 batches | ms/batch 64.67 | loss  0.16567\n",
      "| epoch   4 |  1100/   18 batches | ms/batch 64.69 | loss  0.17458\n",
      "| epoch   4 |  1200/   18 batches | ms/batch 64.68 | loss  0.16869\n",
      "| epoch   4 |  1300/   18 batches | ms/batch 64.81 | loss  0.16942\n",
      "| epoch   4 |  1400/   18 batches | ms/batch 64.59 | loss  0.18366\n",
      "| epoch   4 |  1500/   18 batches | ms/batch 64.47 | loss  0.17354\n",
      "| epoch   4 |  1600/   18 batches | ms/batch 64.48 | loss  0.17506\n",
      "| epoch   4 |  1700/   18 batches | ms/batch 64.48 | loss  0.18240\n",
      "| epoch   4 |  1800/   18 batches | ms/batch 64.62 | loss  0.16480\n",
      "| epoch   4 |  1900/   18 batches | ms/batch 64.46 | loss  0.17851\n",
      "| epoch   4 |  2000/   18 batches | ms/batch 64.44 | loss  0.17113\n",
      "| epoch   4 |  2100/   18 batches | ms/batch 64.37 | loss  0.16156\n",
      "| epoch   4 |  2200/   18 batches | ms/batch 64.34 | loss  0.17112\n",
      "| epoch   4 |  2300/   18 batches | ms/batch 64.94 | loss  0.16650\n",
      "syn mse 0.2151266204885797\n",
      "sen_mse 0.0878093894349699\n",
      "-----------------------------------------------------------------------------------------\n",
      "| epoch   5 |   100/   18 batches | ms/batch 64.15 | loss  0.24138\n",
      "| epoch   5 |   200/   18 batches | ms/batch 64.00 | loss  0.22818\n",
      "| epoch   5 |   300/   18 batches | ms/batch 64.20 | loss  0.23143\n",
      "| epoch   5 |   400/   18 batches | ms/batch 64.30 | loss  0.22878\n",
      "| epoch   5 |   500/   18 batches | ms/batch 64.27 | loss  0.22988\n",
      "| epoch   5 |   600/   18 batches | ms/batch 64.35 | loss  0.22441\n",
      "| epoch   5 |   700/   18 batches | ms/batch 64.53 | loss  0.22529\n",
      "| epoch   5 |   800/   18 batches | ms/batch 64.56 | loss  0.23407\n",
      "| epoch   5 |   900/   18 batches | ms/batch 64.67 | loss  0.22502\n",
      "| epoch   5 |  1000/   18 batches | ms/batch 64.53 | loss  0.22478\n",
      "| epoch   5 |  1100/   18 batches | ms/batch 64.57 | loss  0.22453\n",
      "| epoch   5 |  1200/   18 batches | ms/batch 64.43 | loss  0.22476\n",
      "| epoch   5 |  1300/   18 batches | ms/batch 64.76 | loss  0.22444\n",
      "| epoch   5 |  1400/   18 batches | ms/batch 65.11 | loss  0.22214\n",
      "| epoch   5 |  1500/   18 batches | ms/batch 65.17 | loss  0.22667\n",
      "| epoch   5 |  1600/   18 batches | ms/batch 65.07 | loss  0.21915\n",
      "| epoch   5 |  1700/   18 batches | ms/batch 65.05 | loss  0.23038\n",
      "| epoch   5 |  1800/   18 batches | ms/batch 64.94 | loss  0.22192\n",
      "| epoch   5 |  1900/   18 batches | ms/batch 64.84 | loss  0.22253\n",
      "| epoch   5 |  2000/   18 batches | ms/batch 64.95 | loss  0.22094\n",
      "| epoch   5 |  2100/   18 batches | ms/batch 85.76 | loss  0.22446\n",
      "| epoch   5 |  2200/   18 batches | ms/batch 64.74 | loss  0.22469\n",
      "| epoch   5 |  2300/   18 batches | ms/batch 64.82 | loss  0.21663\n",
      "| epoch   5 |   100/   18 batches | ms/batch 65.70 | loss  0.18750\n",
      "| epoch   5 |   200/   18 batches | ms/batch 64.76 | loss  0.17486\n",
      "| epoch   5 |   300/   18 batches | ms/batch 65.03 | loss  0.18491\n",
      "| epoch   5 |   400/   18 batches | ms/batch 65.13 | loss  0.17137\n",
      "| epoch   5 |   500/   18 batches | ms/batch 64.96 | loss  0.17419\n",
      "| epoch   5 |   600/   18 batches | ms/batch 65.08 | loss  0.17619\n",
      "| epoch   5 |   700/   18 batches | ms/batch 65.07 | loss  0.16497\n",
      "| epoch   5 |   800/   18 batches | ms/batch 65.12 | loss  0.17450\n",
      "| epoch   5 |   900/   18 batches | ms/batch 65.23 | loss  0.16786\n",
      "| epoch   5 |  1000/   18 batches | ms/batch 65.26 | loss  0.17568\n",
      "| epoch   5 |  1100/   18 batches | ms/batch 65.27 | loss  0.16274\n",
      "| epoch   5 |  1200/   18 batches | ms/batch 65.23 | loss  0.17329\n",
      "| epoch   5 |  1300/   18 batches | ms/batch 65.24 | loss  0.16588\n",
      "| epoch   5 |  1400/   18 batches | ms/batch 65.37 | loss  0.16967\n",
      "| epoch   5 |  1500/   18 batches | ms/batch 65.34 | loss  0.15990\n",
      "| epoch   5 |  1600/   18 batches | ms/batch 65.33 | loss  0.16635\n",
      "| epoch   5 |  1700/   18 batches | ms/batch 65.34 | loss  0.16775\n",
      "| epoch   5 |  1800/   18 batches | ms/batch 65.29 | loss  0.17554\n",
      "| epoch   5 |  1900/   18 batches | ms/batch 65.40 | loss  0.17427\n",
      "| epoch   5 |  2000/   18 batches | ms/batch 65.36 | loss  0.17303\n",
      "| epoch   5 |  2100/   18 batches | ms/batch 65.38 | loss  0.16771\n",
      "| epoch   5 |  2200/   18 batches | ms/batch 65.30 | loss  0.16573\n",
      "| epoch   5 |  2300/   18 batches | ms/batch 65.41 | loss  0.15874\n",
      "syn mse 0.20826143481735462\n",
      "sen_mse 0.08820411669175775\n",
      "-----------------------------------------------------------------------------------------\n",
      "| epoch   6 |   100/   18 batches | ms/batch 64.95 | loss  0.22952\n",
      "| epoch   6 |   200/   18 batches | ms/batch 64.02 | loss  0.23040\n",
      "| epoch   6 |   300/   18 batches | ms/batch 64.24 | loss  0.23705\n",
      "| epoch   6 |   400/   18 batches | ms/batch 64.25 | loss  0.21771\n",
      "| epoch   6 |   500/   18 batches | ms/batch 64.27 | loss  0.22849\n",
      "| epoch   6 |   600/   18 batches | ms/batch 64.37 | loss  0.22589\n",
      "| epoch   6 |   700/   18 batches | ms/batch 64.48 | loss  0.21919\n",
      "| epoch   6 |   800/   18 batches | ms/batch 64.53 | loss  0.22072\n",
      "| epoch   6 |   900/   18 batches | ms/batch 64.63 | loss  0.22008\n",
      "| epoch   6 |  1000/   18 batches | ms/batch 64.68 | loss  0.21094\n",
      "| epoch   6 |  1100/   18 batches | ms/batch 65.50 | loss  0.22616\n",
      "| epoch   6 |  1200/   18 batches | ms/batch 65.67 | loss  0.22269\n",
      "| epoch   6 |  1300/   18 batches | ms/batch 65.47 | loss  0.21267\n",
      "| epoch   6 |  1400/   18 batches | ms/batch 64.96 | loss  0.22482\n",
      "| epoch   6 |  1500/   18 batches | ms/batch 64.53 | loss  0.21535\n",
      "| epoch   6 |  1600/   18 batches | ms/batch 64.39 | loss  0.21553\n",
      "| epoch   6 |  1700/   18 batches | ms/batch 64.43 | loss  0.21704\n",
      "| epoch   6 |  1800/   18 batches | ms/batch 64.50 | loss  0.22425\n",
      "| epoch   6 |  1900/   18 batches | ms/batch 64.34 | loss  0.21710\n",
      "| epoch   6 |  2000/   18 batches | ms/batch 64.22 | loss  0.22167\n",
      "| epoch   6 |  2100/   18 batches | ms/batch 64.14 | loss  0.22140\n",
      "| epoch   6 |  2200/   18 batches | ms/batch 89.30 | loss  0.22364\n",
      "| epoch   6 |  2300/   18 batches | ms/batch 63.67 | loss  0.21464\n",
      "| epoch   6 |   100/   18 batches | ms/batch 64.70 | loss  0.17472\n",
      "| epoch   6 |   200/   18 batches | ms/batch 63.98 | loss  0.16648\n",
      "| epoch   6 |   300/   18 batches | ms/batch 64.20 | loss  0.17421\n",
      "| epoch   6 |   400/   18 batches | ms/batch 64.72 | loss  0.16456\n",
      "| epoch   6 |   500/   18 batches | ms/batch 64.82 | loss  0.16902\n",
      "| epoch   6 |   600/   18 batches | ms/batch 64.77 | loss  0.16720\n",
      "| epoch   6 |   700/   18 batches | ms/batch 64.77 | loss  0.17426\n",
      "| epoch   6 |   800/   18 batches | ms/batch 64.77 | loss  0.16860\n",
      "| epoch   6 |   900/   18 batches | ms/batch 64.78 | loss  0.17007\n",
      "| epoch   6 |  1000/   18 batches | ms/batch 64.73 | loss  0.16797\n",
      "| epoch   6 |  1100/   18 batches | ms/batch 64.88 | loss  0.16916\n",
      "| epoch   6 |  1200/   18 batches | ms/batch 64.81 | loss  0.15749\n",
      "| epoch   6 |  1300/   18 batches | ms/batch 64.78 | loss  0.17112\n",
      "| epoch   6 |  1400/   18 batches | ms/batch 64.83 | loss  0.16084\n",
      "| epoch   6 |  1500/   18 batches | ms/batch 64.85 | loss  0.16402\n",
      "| epoch   6 |  1600/   18 batches | ms/batch 64.78 | loss  0.16328\n",
      "| epoch   6 |  1700/   18 batches | ms/batch 64.69 | loss  0.17556\n",
      "| epoch   6 |  1800/   18 batches | ms/batch 64.71 | loss  0.16850\n",
      "| epoch   6 |  1900/   18 batches | ms/batch 64.51 | loss  0.17020\n",
      "| epoch   6 |  2000/   18 batches | ms/batch 64.74 | loss  0.17053\n",
      "| epoch   6 |  2100/   18 batches | ms/batch 64.74 | loss  0.16472\n",
      "| epoch   6 |  2200/   18 batches | ms/batch 64.61 | loss  0.15573\n",
      "| epoch   6 |  2300/   18 batches | ms/batch 64.72 | loss  0.16707\n",
      "syn mse 0.20467287332376113\n",
      "sen_mse 0.08909818069639351\n",
      "-----------------------------------------------------------------------------------------\n",
      "| epoch   7 |   100/   18 batches | ms/batch 64.34 | loss  0.22209\n",
      "| epoch   7 |   200/   18 batches | ms/batch 64.07 | loss  0.21492\n",
      "| epoch   7 |   300/   18 batches | ms/batch 64.25 | loss  0.23058\n",
      "| epoch   7 |   400/   18 batches | ms/batch 64.43 | loss  0.21942\n",
      "| epoch   7 |   500/   18 batches | ms/batch 64.36 | loss  0.22310\n",
      "| epoch   7 |   600/   18 batches | ms/batch 64.54 | loss  0.22438\n",
      "| epoch   7 |   700/   18 batches | ms/batch 64.56 | loss  0.21777\n",
      "| epoch   7 |   800/   18 batches | ms/batch 64.79 | loss  0.22274\n",
      "| epoch   7 |   900/   18 batches | ms/batch 64.74 | loss  0.22066\n",
      "| epoch   7 |  1000/   18 batches | ms/batch 64.70 | loss  0.20632\n",
      "| epoch   7 |  1100/   18 batches | ms/batch 64.57 | loss  0.22030\n",
      "| epoch   7 |  1200/   18 batches | ms/batch 64.69 | loss  0.21776\n",
      "| epoch   7 |  1300/   18 batches | ms/batch 64.85 | loss  0.21882\n",
      "| epoch   7 |  1400/   18 batches | ms/batch 64.72 | loss  0.21903\n",
      "| epoch   7 |  1500/   18 batches | ms/batch 64.65 | loss  0.21824\n",
      "| epoch   7 |  1600/   18 batches | ms/batch 64.47 | loss  0.22060\n",
      "| epoch   7 |  1700/   18 batches | ms/batch 64.29 | loss  0.22209\n",
      "| epoch   7 |  1800/   18 batches | ms/batch 64.39 | loss  0.20859\n",
      "| epoch   7 |  1900/   18 batches | ms/batch 64.29 | loss  0.21463\n",
      "| epoch   7 |  2000/   18 batches | ms/batch 64.29 | loss  0.21234\n",
      "| epoch   7 |  2100/   18 batches | ms/batch 64.19 | loss  0.21409\n",
      "| epoch   7 |  2200/   18 batches | ms/batch 64.24 | loss  0.20553\n",
      "| epoch   7 |  2300/   18 batches | ms/batch 64.31 | loss  0.21529\n",
      "| epoch   7 |   100/   18 batches | ms/batch 65.41 | loss  0.17418\n",
      "| epoch   7 |   200/   18 batches | ms/batch 64.99 | loss  0.16138\n",
      "| epoch   7 |   300/   18 batches | ms/batch 64.94 | loss  0.17174\n",
      "| epoch   7 |   400/   18 batches | ms/batch 64.89 | loss  0.15638\n",
      "| epoch   7 |   500/   18 batches | ms/batch 64.40 | loss  0.17093\n",
      "| epoch   7 |   600/   18 batches | ms/batch 64.09 | loss  0.16861\n",
      "| epoch   7 |   700/   18 batches | ms/batch 63.99 | loss  0.16112\n",
      "| epoch   7 |   800/   18 batches | ms/batch 63.98 | loss  0.16548\n",
      "| epoch   7 |   900/   18 batches | ms/batch 63.85 | loss  0.15994\n",
      "| epoch   7 |  1000/   18 batches | ms/batch 63.88 | loss  0.16216\n",
      "| epoch   7 |  1100/   18 batches | ms/batch 63.90 | loss  0.16483\n",
      "| epoch   7 |  1200/   18 batches | ms/batch 94.57 | loss  0.16532\n",
      "| epoch   7 |  1300/   18 batches | ms/batch 63.35 | loss  0.16592\n",
      "| epoch   7 |  1400/   18 batches | ms/batch 63.61 | loss  0.16508\n",
      "| epoch   7 |  1500/   18 batches | ms/batch 63.57 | loss  0.16535\n",
      "| epoch   7 |  1600/   18 batches | ms/batch 63.64 | loss  0.16169\n",
      "| epoch   7 |  1700/   18 batches | ms/batch 63.78 | loss  0.16137\n",
      "| epoch   7 |  1800/   18 batches | ms/batch 64.53 | loss  0.17068\n",
      "| epoch   7 |  1900/   18 batches | ms/batch 64.67 | loss  0.16147\n",
      "| epoch   7 |  2000/   18 batches | ms/batch 64.51 | loss  0.16897\n",
      "| epoch   7 |  2100/   18 batches | ms/batch 64.63 | loss  0.16873\n",
      "| epoch   7 |  2200/   18 batches | ms/batch 64.44 | loss  0.16589\n",
      "| epoch   7 |  2300/   18 batches | ms/batch 64.71 | loss  0.15991\n",
      "syn mse 0.19934043171855104\n",
      "sen_mse 0.08910459249655135\n",
      "-----------------------------------------------------------------------------------------\n",
      "| epoch   8 |   100/   18 batches | ms/batch 64.41 | loss  0.23189\n",
      "| epoch   8 |   200/   18 batches | ms/batch 64.02 | loss  0.20750\n",
      "| epoch   8 |   300/   18 batches | ms/batch 64.38 | loss  0.21593\n",
      "| epoch   8 |   400/   18 batches | ms/batch 64.36 | loss  0.21536\n",
      "| epoch   8 |   500/   18 batches | ms/batch 64.40 | loss  0.22152\n",
      "| epoch   8 |   600/   18 batches | ms/batch 64.52 | loss  0.21439\n",
      "| epoch   8 |   700/   18 batches | ms/batch 64.66 | loss  0.21437\n",
      "| epoch   8 |   800/   18 batches | ms/batch 64.75 | loss  0.21517\n",
      "| epoch   8 |   900/   18 batches | ms/batch 64.66 | loss  0.21374\n",
      "| epoch   8 |  1000/   18 batches | ms/batch 64.77 | loss  0.21544\n",
      "| epoch   8 |  1100/   18 batches | ms/batch 64.75 | loss  0.20966\n",
      "| epoch   8 |  1200/   18 batches | ms/batch 64.77 | loss  0.21049\n",
      "| epoch   8 |  1300/   18 batches | ms/batch 64.88 | loss  0.21398\n",
      "| epoch   8 |  1400/   18 batches | ms/batch 64.83 | loss  0.21394\n",
      "| epoch   8 |  1500/   18 batches | ms/batch 64.94 | loss  0.21823\n",
      "| epoch   8 |  1600/   18 batches | ms/batch 64.93 | loss  0.21694\n",
      "| epoch   8 |  1700/   18 batches | ms/batch 64.79 | loss  0.21223\n",
      "| epoch   8 |  1800/   18 batches | ms/batch 64.87 | loss  0.21233\n",
      "| epoch   8 |  1900/   18 batches | ms/batch 64.92 | loss  0.20548\n",
      "| epoch   8 |  2000/   18 batches | ms/batch 64.92 | loss  0.20663\n",
      "| epoch   8 |  2100/   18 batches | ms/batch 64.94 | loss  0.21589\n",
      "| epoch   8 |  2200/   18 batches | ms/batch 64.99 | loss  0.21096\n",
      "| epoch   8 |  2300/   18 batches | ms/batch 64.93 | loss  0.20795\n",
      "| epoch   8 |   100/   18 batches | ms/batch 65.88 | loss  0.16566\n",
      "| epoch   8 |   200/   18 batches | ms/batch 65.01 | loss  0.17108\n",
      "| epoch   8 |   300/   18 batches | ms/batch 65.14 | loss  0.15709\n",
      "| epoch   8 |   400/   18 batches | ms/batch 65.90 | loss  0.16469\n",
      "| epoch   8 |   500/   18 batches | ms/batch 65.15 | loss  0.16886\n",
      "| epoch   8 |   600/   18 batches | ms/batch 65.00 | loss  0.16271\n",
      "| epoch   8 |   700/   18 batches | ms/batch 64.32 | loss  0.17054\n",
      "| epoch   8 |   800/   18 batches | ms/batch 64.28 | loss  0.16548\n",
      "| epoch   8 |   900/   18 batches | ms/batch 64.47 | loss  0.16231\n",
      "| epoch   8 |  1000/   18 batches | ms/batch 64.33 | loss  0.16076\n",
      "| epoch   8 |  1100/   18 batches | ms/batch 64.20 | loss  0.16225\n",
      "| epoch   8 |  1200/   18 batches | ms/batch 64.27 | loss  0.16129\n",
      "| epoch   8 |  1300/   18 batches | ms/batch 64.19 | loss  0.15776\n",
      "| epoch   8 |  1400/   18 batches | ms/batch 64.28 | loss  0.16765\n",
      "| epoch   8 |  1500/   18 batches | ms/batch 64.54 | loss  0.15683\n",
      "| epoch   8 |  1600/   18 batches | ms/batch 65.55 | loss  0.16223\n",
      "| epoch   8 |  1700/   18 batches | ms/batch 65.43 | loss  0.15506\n",
      "| epoch   8 |  1800/   18 batches | ms/batch 65.22 | loss  0.15865\n",
      "| epoch   8 |  1900/   18 batches | ms/batch 64.79 | loss  0.16261\n",
      "| epoch   8 |  2000/   18 batches | ms/batch 64.10 | loss  0.16541\n",
      "| epoch   8 |  2100/   18 batches | ms/batch 64.08 | loss  0.15660\n",
      "| epoch   8 |  2200/   18 batches | ms/batch 64.16 | loss  0.16476\n",
      "| epoch   8 |  2300/   18 batches | ms/batch 64.08 | loss  0.16571\n",
      "syn mse 0.18707780562758647\n",
      "sen_mse 0.09010832419014948\n",
      "-----------------------------------------------------------------------------------------\n",
      "| epoch   9 |   100/   18 batches | ms/batch 63.74 | loss  0.22775\n",
      "| epoch   9 |   200/   18 batches | ms/batch 63.31 | loss  0.21153\n",
      "| epoch   9 |   300/   18 batches | ms/batch 63.45 | loss  0.21347\n",
      "| epoch   9 |   400/   18 batches | ms/batch 63.55 | loss  0.21558\n",
      "| epoch   9 |   500/   18 batches | ms/batch 63.55 | loss  0.21624\n",
      "| epoch   9 |   600/   18 batches | ms/batch 64.45 | loss  0.20914\n",
      "| epoch   9 |   700/   18 batches | ms/batch 65.54 | loss  0.21077\n",
      "| epoch   9 |   800/   18 batches | ms/batch 65.23 | loss  0.21292\n",
      "| epoch   9 |   900/   18 batches | ms/batch 65.21 | loss  0.20243\n",
      "| epoch   9 |  1000/   18 batches | ms/batch 65.27 | loss  0.21058\n",
      "| epoch   9 |  1100/   18 batches | ms/batch 102.41 | loss  0.21570\n",
      "| epoch   9 |  1200/   18 batches | ms/batch 64.18 | loss  0.20971\n",
      "| epoch   9 |  1300/   18 batches | ms/batch 64.48 | loss  0.21001\n",
      "| epoch   9 |  1400/   18 batches | ms/batch 64.50 | loss  0.21152\n",
      "| epoch   9 |  1500/   18 batches | ms/batch 64.57 | loss  0.20926\n",
      "| epoch   9 |  1600/   18 batches | ms/batch 64.73 | loss  0.20704\n",
      "| epoch   9 |  1700/   18 batches | ms/batch 64.68 | loss  0.20695\n",
      "| epoch   9 |  1800/   18 batches | ms/batch 64.74 | loss  0.21323\n",
      "| epoch   9 |  1900/   18 batches | ms/batch 64.72 | loss  0.21103\n",
      "| epoch   9 |  2000/   18 batches | ms/batch 64.77 | loss  0.20146\n",
      "| epoch   9 |  2100/   18 batches | ms/batch 64.75 | loss  0.20448\n",
      "| epoch   9 |  2200/   18 batches | ms/batch 64.84 | loss  0.21643\n",
      "| epoch   9 |  2300/   18 batches | ms/batch 64.76 | loss  0.20880\n",
      "| epoch   9 |   100/   18 batches | ms/batch 65.67 | loss  0.16611\n",
      "| epoch   9 |   200/   18 batches | ms/batch 64.97 | loss  0.16688\n",
      "| epoch   9 |   300/   18 batches | ms/batch 64.97 | loss  0.16169\n",
      "| epoch   9 |   400/   18 batches | ms/batch 64.86 | loss  0.15350\n",
      "| epoch   9 |   500/   18 batches | ms/batch 64.81 | loss  0.15500\n",
      "| epoch   9 |   600/   18 batches | ms/batch 64.86 | loss  0.17103\n",
      "| epoch   9 |   700/   18 batches | ms/batch 64.87 | loss  0.15587\n",
      "| epoch   9 |   800/   18 batches | ms/batch 64.85 | loss  0.16219\n",
      "| epoch   9 |   900/   18 batches | ms/batch 64.79 | loss  0.15474\n",
      "| epoch   9 |  1000/   18 batches | ms/batch 64.95 | loss  0.15711\n",
      "| epoch   9 |  1100/   18 batches | ms/batch 65.19 | loss  0.16073\n",
      "| epoch   9 |  1200/   18 batches | ms/batch 65.28 | loss  0.15885\n",
      "| epoch   9 |  1300/   18 batches | ms/batch 65.27 | loss  0.15893\n",
      "| epoch   9 |  1400/   18 batches | ms/batch 65.27 | loss  0.16448\n",
      "| epoch   9 |  1500/   18 batches | ms/batch 64.74 | loss  0.16345\n",
      "| epoch   9 |  1600/   18 batches | ms/batch 64.76 | loss  0.15812\n",
      "| epoch   9 |  1700/   18 batches | ms/batch 64.58 | loss  0.16506\n",
      "| epoch   9 |  1800/   18 batches | ms/batch 64.61 | loss  0.17054\n",
      "| epoch   9 |  1900/   18 batches | ms/batch 64.54 | loss  0.15207\n",
      "| epoch   9 |  2000/   18 batches | ms/batch 64.36 | loss  0.16045\n",
      "| epoch   9 |  2100/   18 batches | ms/batch 63.78 | loss  0.16086\n",
      "| epoch   9 |  2200/   18 batches | ms/batch 63.79 | loss  0.15384\n",
      "| epoch   9 |  2300/   18 batches | ms/batch 63.87 | loss  0.15983\n",
      "syn mse 0.18386133579325392\n",
      "sen_mse 0.08963999723942859\n",
      "-----------------------------------------------------------------------------------------\n",
      "| epoch  10 |   100/   18 batches | ms/batch 63.55 | loss  0.21663\n",
      "| epoch  10 |   200/   18 batches | ms/batch 63.28 | loss  0.21753\n",
      "| epoch  10 |   300/   18 batches | ms/batch 63.98 | loss  0.21139\n",
      "| epoch  10 |   400/   18 batches | ms/batch 64.29 | loss  0.20405\n",
      "| epoch  10 |   500/   18 batches | ms/batch 64.62 | loss  0.21202\n",
      "| epoch  10 |   600/   18 batches | ms/batch 64.63 | loss  0.20875\n",
      "| epoch  10 |   700/   18 batches | ms/batch 64.39 | loss  0.22063\n",
      "| epoch  10 |   800/   18 batches | ms/batch 64.25 | loss  0.20335\n",
      "| epoch  10 |   900/   18 batches | ms/batch 64.39 | loss  0.21105\n",
      "| epoch  10 |  1000/   18 batches | ms/batch 64.35 | loss  0.20642\n",
      "| epoch  10 |  1100/   18 batches | ms/batch 64.33 | loss  0.20260\n",
      "| epoch  10 |  1200/   18 batches | ms/batch 64.49 | loss  0.21925\n",
      "| epoch  10 |  1300/   18 batches | ms/batch 64.39 | loss  0.20763\n",
      "| epoch  10 |  1400/   18 batches | ms/batch 64.52 | loss  0.21103\n",
      "| epoch  10 |  1500/   18 batches | ms/batch 64.77 | loss  0.21300\n",
      "| epoch  10 |  1600/   18 batches | ms/batch 64.54 | loss  0.21577\n",
      "| epoch  10 |  1700/   18 batches | ms/batch 64.68 | loss  0.21236\n",
      "| epoch  10 |  1800/   18 batches | ms/batch 64.58 | loss  0.21292\n",
      "| epoch  10 |  1900/   18 batches | ms/batch 64.58 | loss  0.20708\n",
      "| epoch  10 |  2000/   18 batches | ms/batch 65.20 | loss  0.20761\n",
      "| epoch  10 |  2100/   18 batches | ms/batch 65.11 | loss  0.20885\n",
      "| epoch  10 |  2200/   18 batches | ms/batch 64.97 | loss  0.21189\n",
      "| epoch  10 |  2300/   18 batches | ms/batch 65.00 | loss  0.20778\n",
      "| epoch  10 |   100/   18 batches | ms/batch 65.97 | loss  0.16304\n",
      "| epoch  10 |   200/   18 batches | ms/batch 65.23 | loss  0.16080\n",
      "| epoch  10 |   300/   18 batches | ms/batch 65.37 | loss  0.16057\n",
      "| epoch  10 |   400/   18 batches | ms/batch 65.37 | loss  0.16225\n",
      "| epoch  10 |   500/   18 batches | ms/batch 65.42 | loss  0.15547\n",
      "| epoch  10 |   600/   18 batches | ms/batch 65.36 | loss  0.16745\n",
      "| epoch  10 |   700/   18 batches | ms/batch 65.60 | loss  0.16300\n",
      "| epoch  10 |   800/   18 batches | ms/batch 65.37 | loss  0.16689\n",
      "| epoch  10 |   900/   18 batches | ms/batch 65.39 | loss  0.15724\n",
      "| epoch  10 |  1000/   18 batches | ms/batch 65.35 | loss  0.15699\n",
      "| epoch  10 |  1100/   18 batches | ms/batch 65.44 | loss  0.15230\n",
      "| epoch  10 |  1200/   18 batches | ms/batch 65.28 | loss  0.16267\n",
      "| epoch  10 |  1300/   18 batches | ms/batch 65.35 | loss  0.15448\n",
      "| epoch  10 |  1400/   18 batches | ms/batch 64.78 | loss  0.16518\n",
      "| epoch  10 |  1500/   18 batches | ms/batch 64.75 | loss  0.15430\n",
      "| epoch  10 |  1600/   18 batches | ms/batch 64.68 | loss  0.16270\n",
      "| epoch  10 |  1700/   18 batches | ms/batch 64.81 | loss  0.15425\n",
      "| epoch  10 |  1800/   18 batches | ms/batch 64.81 | loss  0.16837\n",
      "| epoch  10 |  1900/   18 batches | ms/batch 64.81 | loss  0.15849\n",
      "| epoch  10 |  2000/   18 batches | ms/batch 64.64 | loss  0.15316\n",
      "| epoch  10 |  2100/   18 batches | ms/batch 64.74 | loss  0.16070\n",
      "| epoch  10 |  2200/   18 batches | ms/batch 64.78 | loss  0.16036\n",
      "| epoch  10 |  2300/   18 batches | ms/batch 64.70 | loss  0.16282\n",
      "syn mse 0.17977936360953417\n",
      "sen_mse 0.08998357257940166\n",
      "-----------------------------------------------------------------------------------------\n"
     ]
    }
   ],
   "source": [
    "best_val_loss = None\n",
    "try:\n",
    "    for epoch in range(1, epochs+1):\n",
    "        epoch_start_time = time.time()\n",
    "        training(False, train_loader)\n",
    "        training(True, train_loader)\n",
    "        evaluate(test_loader)\n",
    "        print('-'*89)\n",
    "except KeyboardInterrupt:\n",
    "    print('-'*89)\n",
    "    print('Existing from training early')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "metadata": {},
   "outputs": [],
   "source": [
    "model_name = 'id_feat_gene_4'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "metadata": {},
   "outputs": [],
   "source": [
    "#model save\n",
    "torch.save(model.state_dict(), f'{DATA_PATH}/models/{model_name}.p')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Comb(\n",
       "  (drugEncoder): DrugEncoder(\n",
       "    (embed_comp): Embedding(48, 48, padding_idx=0)\n",
       "    (encoderlayer): TransformerEncoderLayer(\n",
       "      (self_attn): MultiheadAttention(\n",
       "        (out_proj): NonDynamicallyQuantizableLinear(in_features=48, out_features=48, bias=True)\n",
       "      )\n",
       "      (linear1): Linear(in_features=48, out_features=2048, bias=True)\n",
       "      (dropout): Dropout(p=0.1, inplace=False)\n",
       "      (linear2): Linear(in_features=2048, out_features=48, bias=True)\n",
       "      (norm1): LayerNorm((48,), eps=1e-05, elementwise_affine=True)\n",
       "      (norm2): LayerNorm((48,), eps=1e-05, elementwise_affine=True)\n",
       "      (dropout1): Dropout(p=0.1, inplace=False)\n",
       "      (dropout2): Dropout(p=0.1, inplace=False)\n",
       "    )\n",
       "    (encoder): TransformerEncoder(\n",
       "      (layers): ModuleList(\n",
       "        (0): TransformerEncoderLayer(\n",
       "          (self_attn): MultiheadAttention(\n",
       "            (out_proj): NonDynamicallyQuantizableLinear(in_features=48, out_features=48, bias=True)\n",
       "          )\n",
       "          (linear1): Linear(in_features=48, out_features=2048, bias=True)\n",
       "          (dropout): Dropout(p=0.1, inplace=False)\n",
       "          (linear2): Linear(in_features=2048, out_features=48, bias=True)\n",
       "          (norm1): LayerNorm((48,), eps=1e-05, elementwise_affine=True)\n",
       "          (norm2): LayerNorm((48,), eps=1e-05, elementwise_affine=True)\n",
       "          (dropout1): Dropout(p=0.1, inplace=False)\n",
       "          (dropout2): Dropout(p=0.1, inplace=False)\n",
       "        )\n",
       "      )\n",
       "    )\n",
       "    (dense_fp): Linear(in_features=167, out_features=32, bias=True)\n",
       "    (dense_gene): Linear(in_features=64, out_features=32, bias=True)\n",
       "    (conv): Conv2d(1, 1, kernel_size=(1, 48), stride=(1, 1))\n",
       "    (FC2): FC2(\n",
       "      (bn): BatchNorm1d(352, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (fc1): Linear(in_features=352, out_features=176, bias=True)\n",
       "      (fc2): Linear(in_features=176, out_features=64, bias=True)\n",
       "      (dropout): Dropout(p=0.3, inplace=False)\n",
       "    )\n",
       "  )\n",
       "  (cellEncoder): CellEncoder(\n",
       "    (embed_ds): Embedding(42, 42, padding_idx=3)\n",
       "    (dense_gene): Linear(in_features=128, out_features=64, bias=True)\n",
       "    (FC2): FC2(\n",
       "      (bn): BatchNorm1d(106, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "      (fc1): Linear(in_features=106, out_features=53, bias=True)\n",
       "      (fc2): Linear(in_features=53, out_features=64, bias=True)\n",
       "      (dropout): Dropout(p=0.3, inplace=False)\n",
       "    )\n",
       "  )\n",
       "  (fc_syn): FC2(\n",
       "    (bn): BatchNorm1d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "    (fc1): Linear(in_features=192, out_features=96, bias=True)\n",
       "    (fc2): Linear(in_features=96, out_features=1, bias=True)\n",
       "    (dropout): Dropout(p=0.3, inplace=False)\n",
       "  )\n",
       "  (fc_ri): FC2(\n",
       "    (bn): BatchNorm1d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "    (fc1): Linear(in_features=128, out_features=64, bias=True)\n",
       "    (fc2): Linear(in_features=64, out_features=1, bias=True)\n",
       "    (dropout): Dropout(p=0.3, inplace=False)\n",
       "  )\n",
       ")"
      ]
     },
     "execution_count": 89,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#model load\n",
    "model.load_state_dict(torch.load(f'{DATA_PATH}/models/{model_name}.p'))\n",
    "model.eval()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### <a id='toc1_4_3_7_'></a>[Evaluation](#toc0_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### <a id='toc1_4_3_7_1_'></a>[Evaluation Metrics](#toc0_)\n",
    "\n",
    "The model's performance is evaluated using several metrics:\n",
    "\n",
    "**Area Under the Receiver Operating Characteristic Curve (AUC-ROC)**\n",
    "\n",
    "This metric evaluates the binary classification performance of the model at various threshold settings. AUC-ROC is a plot of true positive rate against false positive rate at different classification thresholds, and the AUC provides a single scalar value to indicate performance.\n",
    "\n",
    "**Normalized Discounted Cumulative Gain (NDCG)**\n",
    "\n",
    "Used in ranking problems, NDCG measures the performance of ranking predictions based on their relevance. It gives more importance to higher-ranked items, thus accommodating the notion that hits at the top of the ranking are more valuable than hits at lower ranks."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "metadata": {},
   "outputs": [],
   "source": [
    "def dcg_score(y_score, y_true, k):\n",
    "    \"\"\"\n",
    "        https://www.kaggle.com/davidgasquez/ndcg-scorer\n",
    "        y_true: np.array, size= [n_samples]\n",
    "        y_score: np.array, size=[n_samples]\n",
    "        k: int, rank\n",
    "    \"\"\"\n",
    "    order = np.argsort(y_score)[::-1]\n",
    "    y_true = np.take(y_true, order[:k])\n",
    "    \n",
    "    #gain = 2 ** y_true -1\n",
    "    gain = y_true \n",
    "    \n",
    "    discounts = np.log2(np.arange(len(y_true)) + 2)\n",
    "    return np.sum(gain/discounts)\n",
    "\n",
    "def evaluate_accuracy(data_loader):\n",
    "    model.eval()\n",
    "    \n",
    "    syn_all=[]\n",
    "    syn_true_all=[]\n",
    "    ri1_all=[]\n",
    "    ri1_true_all=[]\n",
    "    ri2_all=[]\n",
    "    ri2_true_all=[]\n",
    "    \n",
    "    #loss\n",
    "    with torch.no_grad():\n",
    "        for iteration, sample in enumerate(data_loader):\n",
    "            d1=Variable(sample['d1'])\n",
    "            d1_fp = Variable(sample['d1_fp'].float())\n",
    "            d1_sm = Variable(sample['d1_sm'])\n",
    "            d1_gn = Variable(sample['d1_gn'].float())\n",
    "\n",
    "            d2=Variable(sample['d2'])\n",
    "            d2_fp = Variable(sample['d2_fp'].float())\n",
    "            d2_sm = Variable(sample['d2_sm'])\n",
    "            d2_gn = Variable(sample['d2_gn'].float())\n",
    "\n",
    "            cell = Variable(sample['cell'])\n",
    "            c_ts = Variable(sample['c_ts'])\n",
    "            c_ds = Variable(sample['c_ds'])\n",
    "            c_gn = Variable(sample['c_gn'].float())\n",
    "\n",
    "            syn_true = Variable(sample['syn'].float())\n",
    "            ri_d1=Variable(sample['ri_d1'])\n",
    "            ri_d2=Variable(sample['ri_d2'])\n",
    "\n",
    "\n",
    "            if cuda:\n",
    "                d1=d1.cuda()\n",
    "                d1_fp=d1_fp.cuda()\n",
    "                d1_sm=d1_sm.cuda()\n",
    "                d1_gn=d1_gn.cuda()\n",
    "\n",
    "                d2=d2.cuda()\n",
    "                d2_fp=d2_fp.cuda()\n",
    "                d2_sm=d2_sm.cuda()\n",
    "                d2_gn=d2_gn.cuda()\n",
    "\n",
    "                cell=cell.cuda()\n",
    "                c_ts=c_ts.cuda()\n",
    "                c_ds=c_ds.cuda()\n",
    "                c_gn=c_gn.cuda()\n",
    "\n",
    "\n",
    "            syn,ri1,ri2 = model((d1, d1_fp, d1_sm, d1_gn), (d2, d2_fp, d2_sm, d2_gn), (cell, c_ts,c_ds,c_gn) )\n",
    "            \n",
    "            syn_all.append(syn.data.cpu().numpy())\n",
    "            syn_true_all.append(syn_true.numpy())\n",
    "            \n",
    "            ri1_all.append(ri1.data.cpu().numpy())\n",
    "            ri1_true_all.append(ri_d1.numpy())\n",
    "            \n",
    "            ri2_all.append(ri2.data.cpu().numpy())\n",
    "            ri2_true_all.append(ri_d2.numpy())\n",
    "            \n",
    "    return syn_all, syn_true_all, ri1_all, ri1_true_all, ri2_all, ri2_true_all\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Evaluate synergy prediction"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "metadata": {},
   "outputs": [],
   "source": [
    "syn_all, syn_true_all, ri1_all, ri1_true_all, ri2_all, ri2_true_all= evaluate_accuracy(test_loader)\n",
    "\n",
    "syn_all= [s.item() for syn in syn_all for s in syn]\n",
    "syn_true_all = [s for syn in syn_true_all for s in syn]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 92,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1.0"
      ]
     },
     "execution_count": 92,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#NDCG\n",
    "dcg_score(syn_all,syn_true_all, k=20)/dcg_score(syn_true_all,syn_true_all, k=20)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 93,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.8592733054759554"
      ]
     },
     "execution_count": 93,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#AUPRC\n",
    "metrics.average_precision_score(syn_true_all,  1/(1 + np.exp(-np.array(syn_all))))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 94,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.9637493233172485"
      ]
     },
     "execution_count": 94,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#AUROC\n",
    "metrics.roc_auc_score(syn_true_all,  1/(1 + np.exp(-np.array(syn_all))))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Evaluate sensitivity prediction"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 95,
   "metadata": {},
   "outputs": [],
   "source": [
    "ri1_all= [r.item() for ri in ri1_all for r in ri]\n",
    "ri1_true_all = [r for ri in ri1_true_all for r in ri]\n",
    "\n",
    "ri2_all= [r.item() for ri in ri2_all for r in ri]\n",
    "ri2_true_all = [r for ri in ri2_true_all for r in ri]\n",
    "\n",
    "ri_all=ri1_all+ri2_all\n",
    "ri_true_all=ri1_true_all+ri2_true_all"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 96,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.1481379966891306"
      ]
     },
     "execution_count": 96,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#NDCG\n",
    "dcg_score(ri_all,ri_true_all, k=20)/dcg_score(ri_true_all,ri_true_all, k=20)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.8505135065203695"
      ]
     },
     "execution_count": 97,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#AUC\n",
    "metrics.roc_auc_score(ri_true_all,  1/(1 + np.exp(-np.array(ri_all))))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### <a id='toc1_4_3_8_'></a>[Train Specific Model](#toc0_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "If you want to boost a bit more with general model's test set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 98,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "| epoch  10 |   100/   18 batches | ms/batch 64.32 | loss  0.22585\n",
      "| epoch  10 |   200/   18 batches | ms/batch 63.49 | loss  0.21658\n",
      "| epoch  10 |   300/   18 batches | ms/batch 63.85 | loss  0.21068\n",
      "| epoch  10 |   400/   18 batches | ms/batch 64.02 | loss  0.21123\n",
      "| epoch  10 |   500/   18 batches | ms/batch 64.44 | loss  0.20478\n",
      "| epoch  10 |   100/   18 batches | ms/batch 65.53 | loss  0.17140\n",
      "| epoch  10 |   200/   18 batches | ms/batch 65.00 | loss  0.16609\n",
      "| epoch  10 |   300/   18 batches | ms/batch 65.09 | loss  0.16322\n",
      "| epoch  10 |   400/   18 batches | ms/batch 65.36 | loss  0.16001\n",
      "| epoch  10 |   500/   18 batches | ms/batch 65.37 | loss  0.15451\n"
     ]
    }
   ],
   "source": [
    "#Use major's test set\n",
    "training(False, test_loader)\n",
    "training(True, test_loader)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### <a id='toc1_4_3_8_1_'></a>[Freeze layers](#toc0_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Examine the layer's ID that we'd like to fix or free"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 99,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0 torch.Size([48, 48]) True\n",
      "1 torch.Size([144, 48]) True\n",
      "2 torch.Size([144]) True\n",
      "3 torch.Size([48, 48]) True\n",
      "4 torch.Size([48]) True\n",
      "5 torch.Size([2048, 48]) True\n",
      "6 torch.Size([2048]) True\n",
      "7 torch.Size([48, 2048]) True\n",
      "8 torch.Size([48]) True\n",
      "9 torch.Size([48]) True\n",
      "10 torch.Size([48]) True\n",
      "11 torch.Size([48]) True\n",
      "12 torch.Size([48]) True\n",
      "13 torch.Size([144, 48]) True\n",
      "14 torch.Size([144]) True\n",
      "15 torch.Size([48, 48]) True\n",
      "16 torch.Size([48]) True\n",
      "17 torch.Size([2048, 48]) True\n",
      "18 torch.Size([2048]) True\n",
      "19 torch.Size([48, 2048]) True\n",
      "20 torch.Size([48]) True\n",
      "21 torch.Size([48]) True\n",
      "22 torch.Size([48]) True\n",
      "23 torch.Size([48]) True\n",
      "24 torch.Size([48]) True\n",
      "25 torch.Size([32, 167]) True\n",
      "26 torch.Size([32]) True\n",
      "27 torch.Size([32, 64]) True\n",
      "28 torch.Size([32]) True\n",
      "29 torch.Size([1, 1, 1, 48]) True\n",
      "30 torch.Size([1]) True\n",
      "31 torch.Size([352]) True\n",
      "32 torch.Size([352]) True\n",
      "33 torch.Size([176, 352]) True\n",
      "34 torch.Size([176]) True\n",
      "35 torch.Size([64, 176]) True\n",
      "36 torch.Size([64]) True\n",
      "37 torch.Size([42, 42]) True\n",
      "38 torch.Size([64, 128]) True\n",
      "39 torch.Size([64]) True\n",
      "40 torch.Size([106]) True\n",
      "41 torch.Size([106]) True\n",
      "42 torch.Size([53, 106]) True\n",
      "43 torch.Size([53]) True\n",
      "44 torch.Size([64, 53]) True\n",
      "45 torch.Size([64]) True\n",
      "46 torch.Size([192]) True\n",
      "47 torch.Size([192]) True\n",
      "48 torch.Size([96, 192]) True\n",
      "49 torch.Size([96]) True\n",
      "50 torch.Size([1, 96]) True\n",
      "51 torch.Size([1]) True\n",
      "52 torch.Size([128]) True\n",
      "53 torch.Size([128]) True\n",
      "54 torch.Size([64, 128]) True\n",
      "55 torch.Size([64]) True\n",
      "56 torch.Size([1, 64]) True\n",
      "57 torch.Size([1]) True\n"
     ]
    }
   ],
   "source": [
    "for i, param in enumerate(model.parameters()):\n",
    "    print(i, param.size(), param.requires_grad)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 100,
   "metadata": {},
   "outputs": [],
   "source": [
    "release_after = 46"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 101,
   "metadata": {},
   "outputs": [],
   "source": [
    "for i, param in enumerate(model.parameters()):\n",
    "    if i>=release_after:\n",
    "        param.requires_grad=True\n",
    "    else:\n",
    "        param.requires_grad=False"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Prostate or bone"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 102,
   "metadata": {},
   "outputs": [],
   "source": [
    "# _train_loader_minor=_train_loader_prostate\n",
    "# _test_loader_minor=_test_loader_prostate\n",
    "# _test_minor=_test_prostate\n",
    "_train_loader_minor=_train_loader_bone\n",
    "_test_loader_minor=_test_loader_bone\n",
    "_test_minor=_test_bone"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 103,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "syn mse 0.4604630470275879\n",
      "sen_mse 0.07659931977589925\n",
      "-----------------------------------------------------------------------------------------\n",
      "syn mse 0.43307705720265705\n",
      "sen_mse 0.07582458853721619\n",
      "-----------------------------------------------------------------------------------------\n",
      "syn mse 0.42016200224558514\n",
      "sen_mse 0.07519650459289551\n",
      "-----------------------------------------------------------------------------------------\n",
      "syn mse 0.40270670255025226\n",
      "sen_mse 0.07492810487747192\n",
      "-----------------------------------------------------------------------------------------\n",
      "syn mse 0.39851752916971844\n",
      "sen_mse 0.07496614257494609\n",
      "-----------------------------------------------------------------------------------------\n",
      "syn mse 0.38854416211446124\n",
      "sen_mse 0.07470616698265076\n",
      "-----------------------------------------------------------------------------------------\n",
      "syn mse 0.38684693972269696\n",
      "sen_mse 0.07444736361503601\n",
      "-----------------------------------------------------------------------------------------\n",
      "syn mse 0.38190476099650067\n",
      "sen_mse 0.07440031568209331\n",
      "-----------------------------------------------------------------------------------------\n",
      "syn mse 0.37822500864664715\n",
      "sen_mse 0.07485421498616536\n",
      "-----------------------------------------------------------------------------------------\n",
      "syn mse 0.37540356318155926\n",
      "sen_mse 0.07444271941979726\n",
      "-----------------------------------------------------------------------------------------\n"
     ]
    }
   ],
   "source": [
    "#Use minor's train set\n",
    "try:\n",
    "    for epoch in range(1, epochs+1):\n",
    "        epoch_start_time = time.time()\n",
    "        training(False, _train_loader_minor)\n",
    "        training(True, _train_loader_minor)\n",
    "        evaluate(_test_loader_minor)\n",
    "        print('-'*89)\n",
    "except KeyboardInterrupt:\n",
    "    print('-'*89)\n",
    "    print('Existing from training early')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### <a id='toc1_4_3_9_'></a>[Evaluate Specific Model](#toc0_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Evaluate synergy prediction"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 104,
   "metadata": {},
   "outputs": [],
   "source": [
    "syn_all, syn_true_all, ri1_all, ri1_true_all, ri2_all, ri2_true_all= evaluate_accuracy(_test_loader_minor)\n",
    "\n",
    "syn_all= [s.item() for syn in syn_all for s in syn]\n",
    "syn_true_all = [s for syn in syn_true_all for s in syn]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 105,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.5443161681145564"
      ]
     },
     "execution_count": 105,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#NDCG\n",
    "dcg_score(syn_all,syn_true_all, k=20)/dcg_score(syn_true_all,syn_true_all, k=20)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 106,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.715405562813833"
      ]
     },
     "execution_count": 106,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#AUROC\n",
    "metrics.roc_auc_score(syn_true_all,  1/(1 + np.exp(-np.array(syn_all))))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 107,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.3162988637931727"
      ]
     },
     "execution_count": 107,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#AUPRC\n",
    "metrics.average_precision_score(syn_true_all,  1/(1 + np.exp(-np.array(syn_all))))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Evaluate sensitivity prediction"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 108,
   "metadata": {},
   "outputs": [],
   "source": [
    "ri1_all= [r.item() for ri in ri1_all for r in ri]\n",
    "ri1_true_all = [r for ri in ri1_true_all for r in ri]\n",
    "\n",
    "ri2_all= [r.item() for ri in ri2_all for r in ri]\n",
    "ri2_true_all = [r for ri in ri2_true_all for r in ri]\n",
    "\n",
    "ri_all=ri1_all+ri2_all\n",
    "ri_true_all=ri1_true_all+ri2_true_all"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 109,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.0"
      ]
     },
     "execution_count": 109,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#NDCG\n",
    "dcg_score(ri_all,ri_true_all, k=20)/dcg_score(ri_true_all,ri_true_all, k=20)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 110,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.7249652294853963"
      ]
     },
     "execution_count": 110,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#AUC\n",
    "metrics.roc_auc_score(ri_true_all,  1/(1 + np.exp(-np.array(ri_all))))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### <a id='toc1_4_3_9_1_'></a>[Select the top ranked drug combinations](#toc0_)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 111,
   "metadata": {},
   "outputs": [],
   "source": [
    "syn_all_prob=1/(1 + np.exp(-np.array(syn_all)))\n",
    "order = np.argsort(syn_all_prob)[::-1]\n",
    "syn_true_all_order = np.take(syn_true_all, order[:20])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 112,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "AK-77283 , CHEMBL2144205 , TC-32 , bone , c3716 , 0 , 0 , 0.0\n",
      "DASATINIB , BEZ-235 , TC-32 , bone , c3716 , 0 , 0 , 1.0\n",
      "117048-62-1 , BEZ-235 , TC-32 , bone , c3716 , 0 , 0 , 0.0\n",
      "CARFILZOMIB (PR-171) , CHEMBL2144205 , TC-32 , bone , c3716 , 0 , 0 , 0.0\n",
      "Sepantronium , 658084-64-1 , TC-32 , bone , c3716 , 0 , 0 , 1.0\n",
      "AK-77283 , 658084-64-1 , TC-32 , bone , c3716 , 0 , 0 , 1.0\n",
      "117048-62-1 , 658084-64-1 , TC-32 , bone , c3716 , 0 , 0 , 1.0\n",
      "FOSTAMATINIB (R788) , 658084-64-1 , TC-32 , bone , c3716 , 0 , 0 , 1.0\n",
      "BARDOXOLONE METHYL , 658084-64-1 , TC-32 , bone , c3716 , 0 , 0 , 1.0\n",
      "PALBOCICLIB , 658084-64-1 , TC-32 , bone , c3716 , 0 , 0 , 1.0\n",
      "AK-77283 , CARFILZOMIB (PR-171) , TC-71 , bone , c4817 , 0 , 0 , 0.0\n",
      "AK-77283 , CHEMBL3322515 , TC-32 , bone , c3716 , 0 , 0 , 0.0\n",
      "MEBENDAZOLE , PD325901 , TC-32 , bone , c3716 , 0 , 0 , 1.0\n",
      "117048-62-1 , CHEMBL3322515 , TC-32 , bone , c3716 , 0 , 0 , 0.0\n",
      "ABT-263 (NAVITOCLAX) , 658084-64-1 , TC-32 , bone , c3716 , 0 , 0 , 1.0\n",
      "TRAMETINIB (GSK1120212) , CHEMBL2144205 , TC-32 , bone , c3716 , 0 , 0 , 0.0\n",
      "ISODONOL , PD325901 , TC-32 , bone , c3716 , 0 , 0 , 1.0\n",
      "BMS-754807 , 658084-64-1 , TC-32 , bone , c3716 , 0 , 0 , 1.0\n",
      "CHEMBL2140543 , PD325901 , TC-32 , bone , c3716 , 0 , 0 , 1.0\n",
      "TRAMETINIB (GSK1120212) , CHEMBL3322515 , TC-32 , bone , c3716 , 0 , 0 , 0.0\n"
     ]
    }
   ],
   "source": [
    "for k in range(20):\n",
    "    comb=_test_minor[order[k]]\n",
    "    print(codes['drugs'].idx2item[comb['d1']], ',', \n",
    "          codes['drugs'].idx2item[comb['d2']], ',', \n",
    "          codes['cell'].idx2item[comb['cell']],  ',', \n",
    "          codes['tissue'].idx2item[comb['c_ts']],  ',',\n",
    "          codes['disease'].idx2item[comb['c_ds']], ',', \n",
    "          comb['ri_d1'], ',',\n",
    "          comb['ri_d2'], ',',\n",
    "          syn_true_all_order[k])\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## <a id='toc1_5_'></a>[Results](#toc0_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### <a id='toc1_5_1_'></a>[Drug response prediction accuracy for data-rich tissues](#toc0_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The first experiment involved training the model on a varying number of input features for both cells and drugs. I was unable to refactor the provided code sufficiently to perform the experiments without certain input features, so the results shown in the table are only for the combination of ID, SMILES, and Target Genes. Results shown in Table 1 below:"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div align=\"center\">\n",
    "<table>\n",
    "  <tr>\n",
    "    <th rowspan='2'>Input Features</th>\n",
    "    <th colspan='2'>Monotherapy Sensitivity</th>\n",
    "    <th colspan='2'>Combination Synergy</th>\n",
    "  </tr>\n",
    "  <tr>\n",
    "    <td>NDCG</td>\n",
    "    <td>AUC</td>\n",
    "    <td>NDCG</td>\n",
    "    <td>AUC</td>\n",
    "  </tr>\n",
    "  <tr>\n",
    "    <td>ID+F+G</td>\n",
    "    <td>0.3388</td>\n",
    "    <td>0.8651</td>\n",
    "    <td>1.0000</td>\n",
    "    <td>0.9627</td>\n",
    "  </tr>\n",
    "</table>\n",
    "<p><b>Table 1:</b> Drug response preduction accuracy for data-rich tissues</p>\n",
    "</div>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Compared to the resuls of the original paper, my own results are a bit mixed. On one hand, the accuracy of the model predictions of combination synergy is actually slightly higher than those reported by [Kim et al. (2020)](https://doi.org/10.1101/2020.02.05.932657) by a margin of approximately 0.08 points. Similarly, the NDCG is comparatively higher at 1.00, whereas the original authors reported an NDCG for combination synergy at 0.8377. These seemingly positive results become more perplexing when we consider the monotherapy sensitivity scores, which are much lower for NDCG and AUC by .0847 and 0.6612 points, respectively. Given that I followed the same approach and used their own architecture for the models themselves, these results are striking. Namely the sensitivity score of 0.3388 for NDCG. The only plausible explanation for the discrepancy leads back to the issue with data. Without any guarantee that the restructuring of the new data to align with the dimensionality of the original data was without errors, it's possible that some key data present in the original models was absent in my own analysis. Perhaps even more confusing is the fact that in spite of these likely unintended data exclusions, the model performed even better at predicting the combination synergy. I can only assume that different data and/or preprocessing step(s) were excluded that might offer some insight into the discrepancies between these scores. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### <a id='toc1_5_2_'></a>[Drug response prediction accuracy for bone and prostate cancer](#toc0_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The second experiment shifts the focus of analysis to understudied tissues, namely bone and prostate. To perform this experiment, [Kim et al. (2020)](https://doi.org/10.1101/2020.02.05.932657) transferred the model parameters from data-rich tissues to data-poor tissues, hypothesizing that the model would exhibit improved performance at predicting the synergy and sensitivity for the data-poor tissues given the pretraining. To reproduce this experiment, I followed the same methodology using only the ID+F+G feature model retrained on bone and prostate, respectively. The results can be seen below in Table 2."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div align=\"center\">\n",
    "<table>\n",
    "  <tr>\n",
    "    <th rowspan='2'>Tissue</th>\n",
    "    <th rowspan='2'>Input Features</th>\n",
    "    <th colspan='2'>Monotherapy Sensitivity</th>\n",
    "    <th colspan='2'>Combination Synergy</th>\n",
    "  </tr>\n",
    "  <tr>\n",
    "    <td>NDCG</td>\n",
    "    <td>AUC</td>\n",
    "    <td>NDCG</td>\n",
    "    <td>AUC</td>\n",
    "  </tr>\n",
    "  <tr>\n",
    "    <td>Bone</td>\n",
    "    <td>ID+F+G</td>\n",
    "    <td>0.0000</td>\n",
    "    <td>0.7074</td>\n",
    "    <td>0.3760</td>\n",
    "    <td>0.6655</td>\n",
    "  </tr>\n",
    "  <tr>\n",
    "    <td>Prostate</td>\n",
    "    <td>ID+F+G</td>\n",
    "    <td>0.0897</td>\n",
    "    <td>0.8143</td>\n",
    "    <td>0.9644</td>\n",
    "    <td>0.9690</td>\n",
    "  </tr>\n",
    "</table>\n",
    "<p><b>Table 2:</b> Drug response preduction accuracy for bone and prostate cancer</p>\n",
    "</div>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The results of this experiment were more striking in terms of their deviance from those in the original paper. Namely, the fact that bone received an NDCG score of 0 for monotherapy sensitivity compared with the perfect 1.000 presented in the original paper speaks to the issues in the training data. This is all but confirmed by the fact that scores for combination synergy did not deviate to the same degree. For bone in particular, the NDCG score for combination synergy was 0.3760 and 0.3599 for the reproduction and original paper, respectively. Given I used the same methodology for training the models and computing all of the metrics, the only plausible explanation for the discrepancies ties back to the availability of data for training. Interestingly, prostate did not deviate as significantly from the results in the original paper compared with bone. Granted, the NDCG for monotherapy sensitivity was still much lower at 0.0897 compared to 0.9596, but the AUC for both monotherapy sensitivity and combination synergy outperformed the original results reported in the paper."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### <a id='toc1_5_3_'></a>[Ranking Predicted Synergenistic Drug Combinations](#toc0_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The final experiment conducted in the original paper was a case study in which the authors listed the top 20 predicted drug combinations based on the estimated probability that those combinations were synergistic. Unfortunately, in all the prints of the paper I researched I was unable to find the table in which they referenced their results for this experiment, but they do make a note on page 8 regarding the percentage of the predictions that were actually synergistic as an accuracy metric for both bone and prostate. To reproduce these experiments, I followed the same methodology for prostate and bone, selecting the top 20 predicted drug combinations and offer the results in Tables 3 and 4, respectively."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### <a id='toc1_5_3_1_'></a>[Prostate Top Predictions](#toc0_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div align=\"center\">\n",
    "<table>\n",
    "  <thead>\n",
    "    <tr>\n",
    "      <th>#</th>\n",
    "      <th>Drug 1</th>\n",
    "      <th>Drug 2</th>\n",
    "      <th>Cell</th>\n",
    "      <th>Tissue</th>\n",
    "      <th>Disease</th>\n",
    "      <th>ri_d1</th>\n",
    "      <th>ri_d2</th>\n",
    "      <th>syn_true_all_order</th>\n",
    "    </tr>\n",
    "  </thead>\n",
    "  <tbody>\n",
    "    <tr>\n",
    "      <td>0</td>\n",
    "      <td>AXITINIB</td>\n",
    "      <td>CABAZITAXEL</td>\n",
    "      <td>DU-145</td>\n",
    "      <td>prostate</td>\n",
    "      <td>c4863</td>\n",
    "      <td>0</td>\n",
    "      <td>0</td>\n",
    "      <td>1.0</td>\n",
    "    </tr>\n",
    "    <tr>\n",
    "      <td>1</td>\n",
    "      <td>ACTINOMYCIN D</td>\n",
    "      <td>IXABEPILONE</td>\n",
    "      <td>PC-3</td>\n",
    "      <td>prostate</td>\n",
    "      <td>c4863</td>\n",
    "      <td>0</td>\n",
    "      <td>0</td>\n",
    "      <td>1.0</td>\n",
    "    </tr>\n",
    "    <tr>\n",
    "      <td>2</td>\n",
    "      <td>RUXOLITINIB</td>\n",
    "      <td>CABAZITAXEL</td>\n",
    "      <td>DU-145</td>\n",
    "      <td>prostate</td>\n",
    "      <td>c4863</td>\n",
    "      <td>0</td>\n",
    "      <td>0</td>\n",
    "      <td>1.0</td>\n",
    "    </tr>\n",
    "    <tr>\n",
    "      <td>3</td>\n",
    "      <td>FLUDARABINE BASE</td>\n",
    "      <td>IXABEPILONE</td>\n",
    "      <td>DU-145</td>\n",
    "      <td>prostate</td>\n",
    "      <td>c4863</td>\n",
    "      <td>0</td>\n",
    "      <td>0</td>\n",
    "      <td>1.0</td>\n",
    "    </tr>\n",
    "    <tr>\n",
    "      <td>4</td>\n",
    "      <td>VINBLASTINE SULFATE</td>\n",
    "      <td>IXABEPILONE</td>\n",
    "      <td>PC-3</td>\n",
    "      <td>prostate</td>\n",
    "      <td>c4863</td>\n",
    "      <td>0</td>\n",
    "      <td>0</td>\n",
    "      <td>1.0</td>\n",
    "    </tr>\n",
    "    <tr>\n",
    "      <td>5</td>\n",
    "      <td>AMIFOSTINE</td>\n",
    "      <td>IXABEPILONE</td>\n",
    "      <td>PC-3</td>\n",
    "      <td>prostate</td>\n",
    "      <td>c4863</td>\n",
    "      <td>0</td>\n",
    "      <td>0</td>\n",
    "      <td>1.0</td>\n",
    "    </tr>\n",
    "    <tr>\n",
    "      <td>6</td>\n",
    "      <td>IMATINIB</td>\n",
    "      <td>IXABEPILONE</td>\n",
    "      <td>PC-3</td>\n",
    "      <td>prostate</td>\n",
    "      <td>c4863</td>\n",
    "      <td>0</td>\n",
    "      <td>0</td>\n",
    "      <td>1.0</td>\n",
    "    </tr>\n",
    "    <tr>\n",
    "      <td>7</td>\n",
    "      <td>IMATINIB</td>\n",
    "      <td>IXABEPILONE</td>\n",
    "      <td>DU-145</td>\n",
    "      <td>prostate</td>\n",
    "      <td>c4863</td>\n",
    "      <td>0</td>\n",
    "      <td>0</td>\n",
    "      <td>1.0</td>\n",
    "    </tr>\n",
    "    <tr>\n",
    "      <td>8</td>\n",
    "      <td>TOPOTECAN HYDROCHLORIDE</td>\n",
    "      <td>IXABEPILONE</td>\n",
    "      <td>DU-145</td>\n",
    "      <td>prostate</td>\n",
    "      <td>c4863</td>\n",
    "      <td>0</td>\n",
    "      <td>0</td>\n",
    "      <td>1.0</td>\n",
    "    </tr>\n",
    "    <tr>\n",
    "      <td>9</td>\n",
    "      <td>HYDROXYUREA</td>\n",
    "      <td>IXABEPILONE</td>\n",
    "      <td>PC-3</td>\n",
    "      <td>prostate</td>\n",
    "      <td>c4863</td>\n",
    "      <td>0</td>\n",
    "      <td>0</td>\n",
    "      <td>1.0</td>\n",
    "    </tr>\n",
    "    <tr>\n",
    "      <td>10</td>\n",
    "      <td>TEMOZOLOMIDE</td>\n",
    "      <td>IXABEPILONE</td>\n",
    "      <td>PC-3</td>\n",
    "      <td>prostate</td>\n",
    "      <td>c4863</td>\n",
    "      <td>0</td>\n",
    "      <td>0</td>\n",
    "      <td>1.0</td>\n",
    "    </tr>\n",
    "    <tr>\n",
    "      <td>11</td>\n",
    "      <td>CO-V</td>\n",
    "      <td>IXABEPILONE</td>\n",
    "      <td>DU-145</td>\n",
    "      <td>prostate</td>\n",
    "      <td>c4863</td>\n",
    "      <td>0</td>\n",
    "      <td>0</td>\n",
    "      <td>1.0</td>\n",
    "    </tr>\n",
    "    <tr>\n",
    "      <td>12</td>\n",
    "      <td>CHEMBL17639</td>\n",
    "      <td>IXABEPILONE</td>\n",
    "      <td>DU-145</td>\n",
    "      <td>prostate</td>\n",
    "      <td>c4863</td>\n",
    "      <td>0</td>\n",
    "      <td>0</td>\n",
    "      <td>1.0</td>\n",
    "    </tr>\n",
    "    <tr>\n",
    "      <td>13</td>\n",
    "      <td>PLICAMYCIN</td>\n",
    "      <td>IXABEPILONE</td>\n",
    "      <td>DU-145</td>\n",
    "      <td>prostate</td>\n",
    "      <td>c4863</td>\n",
    "      <td>0</td>\n",
    "      <td>0</td>\n",
    "      <td>1.0</td>\n",
    "    </tr>\n",
    "    <tr>\n",
    "      <td>14</td>\n",
    "      <td>ARSENIC TRIOXIDE</td>\n",
    "      <td>NSC256439</td>\n",
    "      <td>DU-145</td>\n",
    "      <td>prostate</td>\n",
    "      <td>c4863</td>\n",
    "      <td>0</td>\n",
    "      <td>0</td>\n",
    "      <td>0.0</td>\n",
    "    </tr>\n",
    "    <tr>\n",
    "      <td>15</td>\n",
    "      <td>PRALATREXATE</td>\n",
    "      <td>IXABEPILONE</td>\n",
    "      <td>DU-145</td>\n",
    "      <td>prostate</td>\n",
    "      <td>c4863</td>\n",
    "      <td>0</td>\n",
    "      <td>0</td>\n",
    "      <td>1.0</td>\n",
    "    </tr>\n",
    "    <tr>\n",
    "      <td>16</td>\n",
    "      <td>LETROZOLE</td>\n",
    "      <td>IXABEPILONE</td>\n",
    "      <td>DU-145</td>\n",
    "      <td>prostate</td>\n",
    "      <td>c4863</td>\n",
    "      <td>0</td>\n",
    "      <td>0</td>\n",
    "      <td>1.0</td>\n",
    "    </tr>\n",
    "    <tr>\n",
    "      <td>17</td>\n",
    "      <td>RUXOLITINIB</td>\n",
    "      <td>ADM HYDROCHLORIDE</td>\n",
    "      <td>DU-145</td>\n",
    "      <td>prostate</td>\n",
    "      <td>c4863</td>\n",
    "      <td>0</td>\n",
    "      <td>0</td>\n",
    "      <td>1.0</td>\n",
    "    </tr>\n",
    "    <tr>\n",
    "      <td>18</td>\n",
    "      <td>ACTINOMYCIN D</td>\n",
    "      <td>ARSENIC TRIOXIDE</td>\n",
    "      <td>DU-145</td>\n",
    "      <td>prostate</td>\n",
    "      <td>c4863</td>\n",
    "      <td>0</td>\n",
    "      <td>0</td>\n",
    "      <td>1.0</td>\n",
    "    </tr>\n",
    "    <tr>\n",
    "      <td>19</td>\n",
    "      <td>PAZOPANIB HYDROCHLORIDE</td>\n",
    "      <td>CABAZITAXEL</td>\n",
    "      <td>PC-3</td>\n",
    "      <td>prostate</td>\n",
    "      <td>c4863</td>\n",
    "      <td>0</td>\n",
    "      <td>0</td>\n",
    "      <td>1.0</td>\n",
    "    </tr>\n",
    "  </tbody>\n",
    "</table>\n",
    "<p><b>Table 3:</b> Top 20 synergy predictions for prostate</p>\n",
    "</div>\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In Table 3 we can see the top 20 predicted synergistic drug combinations for the prostate tissue. According to [Kim et al. (2020)](https://doi.org/10.1101/2020.02.05.932657), they identified that 10 out of the 20 top predicted drug combinations were actually synergistic (50% hit). Indicating that their model had effectively ranked synergistic drug combinations. In my own results, using the indicator column `syn_true_all_order` as a guide, we can see that the ID+F+G prostate model identified truly synergistic drug combinations for 19/20 drugs (95% hit), outperforming the results reported in the original paper."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### <a id='toc1_5_3_2_'></a>[Bone Top Predictions](#toc0_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div align=\"center\">\n",
    "<table border=\"1\">\n",
    "    <tr>\n",
    "        <th>#</th>\n",
    "        <th>Drug 1</th>\n",
    "        <th>Drug 2</th>\n",
    "        <th>Cell</th>\n",
    "        <th>Tissue</th>\n",
    "        <th>Disease</th>\n",
    "        <th>ri_d1</th>\n",
    "        <th>ri_d2</th>\n",
    "        <th>syn_true_all_order</th>\n",
    "    </tr>\n",
    "    <tr>\n",
    "        <td>0</td>\n",
    "        <td>ISODONOL</td>\n",
    "        <td>PD325901</td>\n",
    "        <td>TC-32</td>\n",
    "        <td>bone</td>\n",
    "        <td>c3716</td>\n",
    "        <td>0</td>\n",
    "        <td>0</td>\n",
    "        <td>1.0</td>\n",
    "    </tr>\n",
    "    <tr>\n",
    "        <td>1</td>\n",
    "        <td>AK-77283</td>\n",
    "        <td>OLAPARIB</td>\n",
    "        <td>TC-71</td>\n",
    "        <td>bone</td>\n",
    "        <td>c4817</td>\n",
    "        <td>0</td>\n",
    "        <td>0</td>\n",
    "        <td>0.0</td>\n",
    "    </tr>\n",
    "    <tr>\n",
    "        <td>2</td>\n",
    "        <td>DASATINIB</td>\n",
    "        <td>BEZ-235</td>\n",
    "        <td>TC-32</td>\n",
    "        <td>bone</td>\n",
    "        <td>c3716</td>\n",
    "        <td>0</td>\n",
    "        <td>0</td>\n",
    "        <td>1.0</td>\n",
    "    </tr>\n",
    "    <tr>\n",
    "        <td>3</td>\n",
    "        <td>TRAMETINIB (GSK1120212)</td>\n",
    "        <td>CUDC-101</td>\n",
    "        <td>TC-32</td>\n",
    "        <td>bone</td>\n",
    "        <td>c3716</td>\n",
    "        <td>0</td>\n",
    "        <td>0</td>\n",
    "        <td>0.0</td>\n",
    "    </tr>\n",
    "    <tr>\n",
    "        <td>4</td>\n",
    "        <td>CUDC-101</td>\n",
    "        <td>PD325901</td>\n",
    "        <td>TC-32</td>\n",
    "        <td>bone</td>\n",
    "        <td>c3716</td>\n",
    "        <td>0</td>\n",
    "        <td>0</td>\n",
    "        <td>1.0</td>\n",
    "    </tr>\n",
    "    <tr>\n",
    "        <td>5</td>\n",
    "        <td>ZINC34894448</td>\n",
    "        <td>thapsigargin</td>\n",
    "        <td>TC-32</td>\n",
    "        <td>bone</td>\n",
    "        <td>c3716</td>\n",
    "        <td>0</td>\n",
    "        <td>0</td>\n",
    "        <td>0.0</td>\n",
    "    </tr>\n",
    "    <tr>\n",
    "        <td>6</td>\n",
    "        <td>ZINC34894448</td>\n",
    "        <td>BARDOXOLONE METHYL</td>\n",
    "        <td>TC-32</td>\n",
    "        <td>bone</td>\n",
    "        <td>c3716</td>\n",
    "        <td>0</td>\n",
    "        <td>0</td>\n",
    "        <td>0.0</td>\n",
    "    </tr>\n",
    "    <tr>\n",
    "        <td>7</td>\n",
    "        <td>117048-62-1</td>\n",
    "        <td>CARFILZOMIB (PR-171)</td>\n",
    "        <td>TC-32</td>\n",
    "        <td>bone</td>\n",
    "        <td>c3716</td>\n",
    "        <td>0</td>\n",
    "        <td>0</td>\n",
    "        <td>0.0</td>\n",
    "    </tr>\n",
    "    <tr>\n",
    "        <td>8</td>\n",
    "        <td>117048-62-1</td>\n",
    "        <td>Luminespib</td>\n",
    "        <td>TC-32</td>\n",
    "        <td>bone</td>\n",
    "        <td>c3716</td>\n",
    "        <td>0</td>\n",
    "        <td>0</td>\n",
    "        <td>0.0</td>\n",
    "    </tr>\n",
    "    <tr>\n",
    "        <td>9</td>\n",
    "        <td>ISODONOL</td>\n",
    "        <td>CUDC-101</td>\n",
    "        <td>TC-32</td>\n",
    "        <td>bone</td>\n",
    "        <td>c3716</td>\n",
    "        <td>0</td>\n",
    "        <td>0</td>\n",
    "        <td>0.0</td>\n",
    "    </tr>\n",
    "    <tr>\n",
    "        <td>10</td>\n",
    "        <td>AK-77283</td>\n",
    "        <td>CARFILZOMIB (PR-171)</td>\n",
    "        <td>TC-71</td>\n",
    "        <td>bone</td>\n",
    "        <td>c4817</td>\n",
    "        <td>0</td>\n",
    "        <td>0</td>\n",
    "        <td>0.0</td>\n",
    "    </tr>\n",
    "    <tr>\n",
    "        <td>11</td>\n",
    "        <td>Fingolimod</td>\n",
    "        <td>proscillaridin</td>\n",
    "        <td>TC-32</td>\n",
    "        <td>bone</td>\n",
    "        <td>c3716</td>\n",
    "        <td>0</td>\n",
    "        <td>0</td>\n",
    "        <td>1.0</td>\n",
    "    </tr>\n",
    "    <tr>\n",
    "        <td>12</td>\n",
    "        <td>CHEMBL2140543</td>\n",
    "        <td>PD325901</td>\n",
    "        <td>TC-32</td>\n",
    "        <td>bone</td>\n",
    "        <td>c3716</td>\n",
    "        <td>0</td>\n",
    "        <td>0</td>\n",
    "        <td>1.0</td>\n",
    "    </tr>\n",
    "    <tr>\n",
    "        <td>13</td>\n",
    "        <td>BEZ-235</td>\n",
    "        <td>thapsigargin</td>\n",
    "        <td>TC-32</td>\n",
    "        <td>bone</td>\n",
    "        <td>c3716</td>\n",
    "        <td>0</td>\n",
    "        <td>0</td>\n",
    "        <td>0.0</td>\n",
    "    </tr>\n",
    "    <tr>\n",
    "        <td>14</td>\n",
    "        <td>CHEMBL2140543</td>\n",
    "        <td>BARDOXOLONE METHYL</td>\n",
    "        <td>TC-32</td>\n",
    "        <td>bone</td>\n",
    "        <td>c3716</td>\n",
    "        <td>0</td>\n",
    "        <td>0</td>\n",
    "        <td>0.0</td>\n",
    "    </tr>\n",
    "    <tr>\n",
    "        <td>15</td>\n",
    "        <td>CARFILZOMIB (PR-171)</td>\n",
    "        <td>PANOBINOSTAT</td>\n",
    "        <td>TC-32</td>\n",
    "        <td>bone</td>\n",
    "        <td>c3716</td>\n",
    "        <td>0</td>\n",
    "        <td>0</td>\n",
    "        <td>0.0</td>\n",
    "    </tr>\n",
    "    <tr>\n",
    "        <td>16</td>\n",
    "        <td>117048-62-1</td>\n",
    "        <td>CUDC-101</td>\n",
    "        <td>TC-32</td>\n",
    "        <td>bone</td>\n",
    "        <td>c3716</td>\n",
    "        <td>0</td>\n",
    "        <td>0</td>\n",
    "        <td>0.0</td>\n",
    "    </tr>\n",
    "    <tr>\n",
    "        <td>17</td>\n",
    "        <td>DORSOMORPHIN</td>\n",
    "        <td>thapsigargin</td>\n",
    "        <td>TC-32</td>\n",
    "        <td>bone</td>\n",
    "        <td>c3716</td>\n",
    "        <td>0</td>\n",
    "        <td>0</td>\n",
    "        <td>0.0</td>\n",
    "    </tr>\n",
    "    <tr>\n",
    "        <td>18</td>\n",
    "        <td>BEZ-235</td>\n",
    "        <td>CUDC-101</td>\n",
    "        <td>TC-32</td>\n",
    "        <td>bone</td>\n",
    "        <td>c3716</td>\n",
    "        <td>0</td>\n",
    "        <td>0</td>\n",
    "        <td>0.0</td>\n",
    "    </tr>\n",
    "    <tr>\n",
    "        <td>19</td>\n",
    "        <td>117048-62-1</td>\n",
    "        <td>TRAMETINIB (GSK1120212)</td>\n",
    "        <td>TC-32</td>\n",
    "        <td>bone</td>\n",
    "        <td>c3716</td>\n",
    "        <td>0</td>\n",
    "        <td>0</td>\n",
    "        <td>1.0</td>\n",
    "    </tr>\n",
    "</table>\n",
    "<p><b>Table 4: </b>Top 20 synergy predictions for bone</p>\n",
    "</div>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In the original paper, [Kim et al. (2020)](https://doi.org/10.1101/2020.02.05.932657) reported that their model successfully identified 5/20 true synergistic drug combinations (25% hit); and as can be seen in Table 4, my own results were similar at 6/20 (30% hit). Interestingly, these results seem to align closely with the overall prediction accuracy for prostate being superior to that of bone. As in the previous experiment we saw a similar case in which bone, in comparison to the results in the origianl paper, either stagnated or stayed relatively the same, whereas my own results seem to outperform for prostate."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## <a id='toc1_6_'></a>[Discussion](#toc0_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### <a id='toc1_6_1_'></a>[Issues Encountered](#toc0_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In spite of my best efforts, I was unable to fully reproduce the results of [Kim et al. (2020).](https://doi.org/10.1101/2020.02.05.932657) I do not believe that this paper is reproducible in its current state, given the lack of documentation and the unavailability of data. I believe that the primary issue with this reproduction attempt was the ambiguity of where the data came from, and in what form. While I initially mistook the publicly available data as a good sign that this step would be comparatively easy, I soon found that acquiring the data originally used for this paper was a non-trivial task—as the data available in DrugComb has since been updated to a point in which there was no way to get the names of tissues and diseases aligned with their respective ids, and the data accessed from DrugBank was not documented anywhere in the original paper or in the repository. To that point, the only way I managed to access *some* of the original data was by reaching out to several of the original authors directly; even then, they did not share the complete set of original data they used, which left me struggling to refactor the code to align with the dimensionality and nuances of the combined new and original datasets. For more information on the specific challenges faced from a data perspective, please see the [Scope of Reproducability](#scope-of-reproducability) section above. \n",
    "\n",
    "The challenge of data collection proved to be multifaceted, as the scripts themselves were designed with only the original datasets in mind; this necessitated that I make significant changes throughout the code to align with synthesis of new and original data. I spent a significant amount of time attempting to refactor the code, but without a background in molecular biology and no documentation I was left struggling to interpret whether the dimensions in the new data were in fact aligned with those of the original. Given the number of variables, I was, in every case, forced to consider multiple scenarios when encountering errors running the code; these could have come from a misalignment in the dimensions of the new and original data, an oversight in my code with respect to the methodology used by the authors, or simply a nuance in the functionality of the packages due to differences in versions used. Unfortunately, the lack of documentation precluded me from being able to easily narrow down where the issues were coming from, and I was instead left spending a significant amount of time refactoring the data and code."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### <a id='toc1_6_2_'></a>[Reproducability and Future Considerations](#toc0_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "To reiterate, I do not believe that this paper is reproducible in its current state. Without proper documentation or processes in place for accessing the original data used, there is no way to ensure consistency across reproduction attempts. Furthermore, repeatedly contacting the authors in the hopes of acquiring the original data has diminishing returns as they are likely busy people to the extent that a multitude of students contacting them all requesting the same thing most likely comes across as annoying. On the other hand, it behooves the authors of any academic paper to make their results reproducible and documented—and the persistent issue of requests concerning the data is one created by their own design. This is all to say that there is nuance to be considered with regard to this topic. To what extent is the inability to reproduce a paper the fault of the person attempting the reproduction or the fault of the authors? The reality is most likely somewhere in the middle. Reflecting on my own faults in this process, the late change of paper undoubtedly caused me to have a shorter runway for completing this project. It's also possible that I made a wrong turn in attempting to reuse the authors' code and data. While this approach seemed like the most logical thing to do at the time, it's true that with the time spent refactoring may have been better spent simply building a solution from the ground up. Unfortunately, that was not the avenue I pursued, and the results of this reproduction attempt speak to that.\n",
    "\n",
    "In terms of recommendations to Kim et al. for the future, I would strongly suggest they make the original data (all of it) accessible for reproduction attempts. Sharing only partials creates even more confusion in attempting to align dimensionality of new and original data, as described above. Doing this would narrow the margin of error in reproduction attempts insofaras errors in the code would be significantly easier to troubleshoot. Furthermore, this would cut down on the number of requests the authors receive concerning this, which is undoubtedly in their best interest."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## <a id='toc1_7_'></a>[References](#toc0_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "[1]: Chen, X. (2002). TTD: Therapeutic target database. Nucleic Acids Research, 30(1), 412–415. https://doi.org/10.1093/nar/30.1.412 \n",
    "\n",
    "[2]: HMS LINCS (Harvard Medical School). LINCS MCF 10A Common Project: Rolling-timepoint sensitivity measures of the MCF 10A breast cell line to 8 small molecule perturbagens. Dataset 8 of 15: End-point dose-response metrics for biological replicate 2. http://lincsportal.ccs.miami.edu/datasets/#/view/LDS-1415. Published 2017.\n",
    "\n",
    "[3]: Kim, Y., Zheng, S., Tang, J., Zheng, W. J., Li, Z., & Jiang, X. (2020). Anti-Cancer Drug Synergy Prediction in Understudied Tissues Using Transfer Learning. https://doi.org/10.1101/2020.02.05.932657 \n",
    "\n",
    "[4]: Liu, S., Nie, W., Wang, C., Lu, J., Qiao, Z., Liu, L., Tang, J., Xiao, C., & Anandkumar, A. (2023). Multi-modal Molecule structure–text model for text-based retrieval and editing. Nature Machine Intelligence, 5(12), 1447–1457. https://doi.org/10.1038/s42256-023-00759-6\n",
    "\n",
    "[5]: Wishart, D. S., Knox, C., Guo, A. C., Cheng, D., Shrivastava, S., Tzur, D., Gautam, B., & Hassanali, M. (2007). Drugbank: A Knowledgebase for drugs, drug actions and drug targets. Nucleic Acids Research, 36(suppl_1). https://doi.org/10.1093/nar/gkm958 \n",
    "\n",
    "[6]: Zagidullin, B., Aldahdooh, J., Zheng, S., Wang, W., Wang, Y., Saad, J., Malyutina, A., Jafari, M., Tanoli, Z., Pessia, A., & Tang, J. (2019). Drugcomb: An integrative cancer drug combination data portal. Nucleic Acids Research, 47(W1). https://doi.org/10.1093/nar/gkz337 \n",
    "\n"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "synergy",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.19"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}