Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

example code in libmolgrid github #113

Open
croshong opened this issue Apr 17, 2023 · 4 comments
Open

example code in libmolgrid github #113

croshong opened this issue Apr 17, 2023 · 4 comments

Comments

@croshong
Copy link

I'm now trying to reproduce 3D-cnn result with keras/tensorflow using cross dock dataset
can I regard the example code in https://gnina.github.io/libmolgrid/tutorials/train_basic_CNN_with_Tensorflow.html

as def2017 model generation?

If it's true, to make another model such as HiRes Pose, HiRes Affinity, Def2018
the only thing that I need to do is modify create_model function in example code ?

Thanks in advance

@dkoes
Copy link
Contributor

dkoes commented Apr 17, 2023

Yes, what problems are you running into that would make you think otherwise?

@croshong
Copy link
Author

Thanks for your reply

I could find the model detail in https://github.com/gnina/models

Now I'm trying to make a simple 3D-cnn script with model in https://github.com/gnina/models in Keras/tensorflow

I can modify create model function

The thing that I'm not fully confident is the input generation part which involve molgrid.

In the example script in gnina.github.io/libmolgrid

The input data seems to be splitted into batch which has a size of 50 and fed into model

500 times( epoch) with next_batch() statement

I want to setup train/validation procedure and finally test procedure

to do that for each epoch

I extracted 10 bacth( maybe random?) for training and 10 batch( maybe random?) for validation
and run train_on_batch and test_on_batch

and finally for test,
extract 10 batch and predict with generated model and calculate R^2

attached files are the simple python code which does 3D cnn with keras/tensorlfow made by me

Can you please check whether there is problem or not?

really thanks for your hlep
tt3_v2.txt

@dkoes
Copy link
Contributor

dkoes commented Apr 18, 2023

Why do you think there might be a problem? Have you run the code? What is the output?

@croshong
Copy link
Author

No there is no problem in running example code but I think there is only training data and training process in the example code
so to include validation/test process for my purpose, I have modified the example code as attached file

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants