proGENTRL

In this Blogpost I would like to talk about one of my projects I am currently working on Pro Generative Tensorial Reinforcement Learning AKA proGENTRL. This is a Pytorch Lightning implementation of insilo medicine’s GENTRL.

Recently Pytorch Lightning has gained much popularity due to its lightweight framework and flexibility with handling multiple devices from a single GPU to a HPC cluster. So in order to maximize the efficiency of using GENTRL on multi-GPU environment I have implemented this pytorch lightning code.

Pro Generative Tensorial Reinforcement Learning (proGENTRL)

Installation

Step 1 :

Make a new conda environment and install RDKit.

conda create -c rdkit -n my-rdkit-env rdkit

Then activate this new environment.

conda activate my-rdkit-env

Note : Make sure that the python3 version is 3.5 or higher and pip3 is installed

Step 2 :

Inside this environment install proGENTRL.

git clone https://github.com/Bibyutatsu/proGENTRL.git
cd proGENTRL
python3 setup.py install

Step 3 :

Then I will suggest you install pytorch’s latest version according to your cuda version (For e.g. 10.2)

python3 -m pip uninstall torch torchvision
conda install pytorch torchvision cudatoolkit=10.2 -c pytorch

Step 4 : (Optional)

Making a new Kernel for jupyter notebook is recommended. For making a new kernel please follow these steps.

python3 -m pip install ipykernel
python3 -m ipykernel install --user --name rdkit_kernel

Now when you open jupyter notebook. Go to Change Kernel > rdkit_kernel

With these the installation is over and now we are ready to run the Example.ipynb notebook provided in the Repo.

Usage

I have summarised the basic pipeline for using this module in Example.ipynb. This notebook is a good starting point to look at how proGENTRL works. It incorporates all the steps from VAE Training to Reinforcement Learning and finally Sampling or generation of new molecules in the form of SMILES strings.

The basic flow of the model is: VAE Train :arrow_right: Reinforcement Learning :arrow_right: Sampling

Examples

Below we show examples of generated molecules (more samples here). You can find more explanations in my originarlly forked Repo of GENTRL please visit here

Supporting Information for the paper “Deep learning enables rapid identification of potent DDR1 kinase inhibitors“. Original Repo