NanoAnalysis

NanoAnalysis is a large scale, ML-driven analysis of inorganic nanoparticles in pre-clinical cancer research.

For more information, please refer to the associated publication: https://www.nature.com/articles/s41565-024-01673-7

If you use this data or code, please kindly cite Mendes & Zhang et al. Nature Nanotechnology 19, 867–878 (2024)

Dependencies

Supervised machine learning runs in Python 3.9 using algorithms from scikit-learn and XGBoost. Explainable AI is dependent on SHAP analysis. tqdm is a useful tool to visually track your job progress. A fresh conda environment can be set up using

conda create -n NanoAna python=3.9 pandas
conda activate NanoAna
conda install scikit-learn
conda install -c conda-forge py-xgboost
conda install -c conda-forge shap
conda install tqdm

Alternatively, users could implement the analysis on cloud-based platforms with pre-configured Python environment, e.g. Google Colab, and required packages can be installed using

!pip install xgboost
!pip install shap

Text mining runs in RStudio Version 2022.12.0+353 using functions from tidyverse, quanteda, tm, ggplot2, ggrepel, and cowplot. All libraries can be installed in Rstudio in the console using

install.packages(c("tidyverse","quanteda","quanteda.textmodels","quanteda.textplots","quanteda.textstats","ggplot2","ggrepel","tm","cowplot"))

Descriptions of folders and files

data

The available data sources include:

• Curated dataset from the literature
• Transformed database ready for machine learning training and analysis
• Dataset of publications from Pubmed and Web of Science general searches on inorganic cancer nanoparticles
• Dataset of publications curated by the research team
• Data on materials and cancer types commonly associated with the inorganic cancer nanoparticle space

code

• This folder includes core functions that underlie the analysis pipeline, executable examples for users to run, and an interactive notebook to run text mining analysis.