Reproducibility Archive for Machine Learning-Based Virtual Screening and Molecular Dynamics Simulations for GSK3β Inhibitors in Alzheimer's Disease

This Zenodo repository contains the complete reproducibility archive for the study: “Machine Learning-Based Virtual Screening and Molecular Dynamics Simulations for GSK3β Inhibitors in Alzheimer’s Disease” The archive includes: Machine Learning · Raw and processed datasets · Full descriptor matrices (normalized + raw) · Pre-processing, feature selection, and PCA scripts · Hyperparameter tuning and model development notebooks Virtual Screening · ASINEX screening datasets (full, AD-filtered, ML-filtered) · Screening workflow notebooks and scripts Molecular Docking · Receptor files (cleaned PDB, PDBQT) · Ligand structures (SDF) · Grid parameter files- config file · Docking logs and best poses for selected ligands · Docking preparation and analysis notebooks · Binding energy results from docking Molecular Dynamics (AMBER + GROMACS Hybrid Workflow) · All AMBER input files: tleap.in, min.in, heat.in, density.in, equil.in, mmpbsa.in · GROMACS production input (prod.mdp), GRO/TOP/TPR files · Ligand parameterization files (mol2, frcmod) · Conversion scripts (acpype, parmed) · Per-ligand MD directories with input files · RMSD, RMSF, SASA, Rg, PCA, HBond, DSSP analysis scripts and outputs · Per-residue MMGBSA and MMPBSA energy results Webtool · Full Streamlit webtool implementation · Final trained RF and ET models (.pkl) · Applicability domain statistics (means, covariance matrices, cutoffs) · ML models, AD cutoffs · Validation datasets and codes This archive is provided to ensure complete transparency and computational reproducibility for peer review.It will be made public upon acceptance of the manuscript. First author: Ajwin Joseph MartinCorresponding author: Dr. Dileep Kumar