DrugAutoML 0.0.1

DrugAutoML: An Open-Source Automated Machine Learning and Statistical Evaluation Tool for Bioactivity Prediction in Drug Discovery

DrugAutoML

DrugAutoML is an open-source automated machine learning (AutoML) and statistical evaluation tool designed for bioactivity prediction in drug discovery. It provides data preprocessing, fingerprint generation, model selection, and result analysis pipelines to streamline the machine learning workflow.

Key Features

• Data Preprocessing
  Cleans and validates SMILES, classifies compounds as active/inactive based on IC50 thresholds.

• Fingerprint Calculation
  Generates Morgan (ECFP) fingerprints for molecular representation.

• AutoML Pipeline
  Automates model training with hyperparameter optimization for multiple ML algorithms.

• Model Selection & Analysis
  Provides cross-validation metrics, ROC/PRC curves, confusion matrices, and SHAP explanations.