Rate My Molecule:

Software package for predicting synthetic accesibility of organic molecules. The tools are deployed as a Flask webapp, using PostgreSQL database.

This code was developed by Steven Bennett, while undergoing my PhD in the Jelfs group, jelfs-group.org.

Installation

Prerequisites

• Install Docker OS specific instructions..
• Install docker-compose.

Building the RateMyMolecule Image

The RMM image can be built using the Dockerfile in the repository flask/Dockerfile. This will build the Flask image to the local machine, however, database-based operations will be unable to be run, if the PostgreSQL container is not running.

Database Management

The docker-compose.yml contains the configuation to deploy the full-stack. This does not require that the RMM Flask image is has been built. To build and run the initial database run

    docker-compose build database
    docker-compose up -d database

This will run the database Docker image in the background, creating a user, example, and database example also. It will populate both of them with the SMILES molecules present.

Running the Full Application

To run the full-stack run

    docker-compose build
    docker-compose up

This will run the three images required to serve the website on localhost, database, flask and nginx.

If using this for production, specify a username and password for the PostgresSQL database in the database.conf folder and remove the default secret key from the Dockerfile.

Customising the Application

The SMILES strings displayed by default can be customised. These are located in the init.sql file in databases, and will automatically add to the database. Additionally, this website can be used for any type of binary classification problem as the question can be changed. This can be customised in flask/main/index.html.