Deep reinforcement learning for de novo drug design: a ReLeaSe method execution on a Docker Environment

Citation

This work runs the developed ReLeaSE method on a Docker Environment. Further details about the method can be found in this paper: Mariya Popova, Olexandr Isayev, Alexander Tropsha. Deep Reinforcement Learning for de-novo Drug Design. Science Advances, 2018, Vol. 4, no. 7, eaap7885. DOI: 10.1126/sciadv.aap7885

Tutorial Structure

• Installation
  • Prerequisites
  • Repository
  • Environment Variables
  • Build
  • Run Docker Services

Installation

Prerequisites

• Linux OS or WSL 2 on a Windows 10 (or higher) machine. If you use WSL 2, follow the official guide.
• A Modern NVIDIA GPU, compatible with CUDA 11.3.
• Docker and Docker Compose (Application containers engine). Install it from here.
• The Docker GPU Support enabled on the machine; check it out here.
• The Nvidia Container Toolkit. Install it from here.

Note that you do not need to install the CUDA Toolkit on the host system.

Note that you have to install the NVIDIA drivers on your system.

Repository

Clone the repository:

$ git clone https://github.com/IvanBuccella/SF2Bio
$ cd docker

Environment Variables

Set your own environment variables (the jupyter notebook port) by using the .env-sample file. You can just duplicate and rename it in .env.

Build

Build the local environment with Docker:

$ docker-compose build

Run Docker Services

$ docker-compose up

Enjoy :-)

You can execute an example that uses method by visiting the http://localhost:${HTTP_PORT}/LogP_optimization_demo.ipynb URL.