Potency prediction for Matched Molecular Pairs
Repository for the paper: Anatomy of Potency Predictions Focusing on Structural Analogues with Increasing Potency Differences Including Activity Cliffs
Anaconda can be used to install the .yml file provided.
- regression_models_mmps.ipynb - generates regression models with a 50/50% random and stratified splits for MMP datasets. (Run before SHAP notebook)
- regression_models_shap_mmps.ipynb - generates SV/SHAP values for SVR and RFR for MMP test sets.
- data_final_analysis.ipynb - derives analysis for the computed results, Fig 2, 3 and 4.
- shap_bit_analysis.ipynb - derives SV/SHAP analysis, Fig 5.
- shap_mapping.ipynb - Map SV/SHAP values to test compounds.
- ML - contains Python scripts to support model building and data analysis.
- dataset - contains the dataset used in the analysis.
- ccrlib_master - constains scripts to generate MMP datasets.