This code will be used for comparing the results of machine learning rescoring functions, such as nnscore and rfscore against the results of AutoDock Vina. Decoy compounds from the DUD-E dataset are docked to their receptors using Vina and the ODDT library, and poses are re-scored.
This repository contains:
- A Python script for docking multiple compounds using ODDT: https://oddt.readthedocs.io/en/0.1.1/rst/oddt.html.
- A SLURM bash file for docking to multiple receptors in an HPC environment.
- A Jupyter notebook for generating ROC curves:
ODDT DUDE-E ROC curves.ipynb
The scripts here are tested on two receptors, abl1
and aa2r
.