ML-for-LSD1

This repository contains original codes and data in Machine-Learning-Enabled Virtual Screening For Inhibitors of Lysine-Specific Histone Demethylase 1.

Requirements

• Python 3.7
• Numpy 1.19.2
• Pandas 1.1.3
• RDKit 2020.09.1.0
• scikit-learn 0.23.2
• matplotlib 3.3.2
• torch 1.9.0
• PyTorch Lightning 1.4.9

Dataset for screening

The screening dataset can be found in https://zinc.docking.org/substances/subsets/in-vitro/ . Please add the in-vitro.csv file to ./ML-for-LSD1/screening_base/in-vitro_zinc/ directory.

Script Description

Generate datasets for algorithms from the original pChEMBL dataset.

data_cleansing.py and dataset_construction.py

Hyperparameter Optimization

optimization.py

Algorithm Fitting with Best Performing Hyperparameter Combinations

fitting.py

Some Analysis

plot_learning_curves.py and tsne.py

Neural Network

neural network.py

Virtual Screening

deploy.py and deploy2.py

Utils

utils.py, rdkit_utils.py and nn_utils.py

Acknowledgement

I would like to thank Miss Yufan Liu from University of Surrey for her contribution in code validation and visualization.