This repo was created for a school project. It is an implementation of a very simple classification and QSAR modelling tool.

WARNING: The code is mostly ugly :)

Each module has params.py file where the parameters of the module can be specified. You will probably want to configure at least the datageneration module, which contains definitions of file paths and some basic parameters such as the protein target, clustering threshold...

Please note that you will have to download the decoys as an .sdf file and point to the proper loacation via datamanipulation.params.DECOYS_SDF_FILE_PATH. Actives should be downloaded automatically.

The modules are loaded from the main.py file in the root of the repo. The main.py file includes a method that builds models, a prediction method for pIC₅₀ predicitons and the main method. The main method includes a simple usage example.