Using Machine Learning to Examine Pre-Transplant Factors Influencing De novo HLA-Specific Antibody Development Post-Kidney Transplant

code contains the scripts to run processing

kidney_tx_analysis.sh submits the following processing on SLURM:

• prep_data.py: basic data cleaning
• associations_de_novo.py: generates associations using dython. This is included for completeness, other_plots.py is used to generate the publication figures
• pre_dt_rf_cat_xgb.py: trains and tests CART, RF, Catboost and XGBoost classifiers
• best_results.py: prints the F1 scores of classifier

shap_top_3_models.py: run SHAP and generate figures for the top 3 best performing models simple_feature_importance.py: calculate feature importance for the top 3 best performing models and generate figures other_plots.py: generate all other figures, including for exploratory data analysis.