
Explainable QSAR & SHAP Molecular Descriptors
Descriptor-level feature attribution using SHAP values to explain bioactivity drivers in non-linear tree-based QSAR models.

Curated portfolio of academic research investigations, independent computational studies, and bioinformatics coursework across machine learning, transcriptomics, and drug discovery.

Descriptor-level feature attribution using SHAP values to explain bioactivity drivers in non-linear tree-based QSAR models.

Machine learning screening workflow predicting molecular safety, toxicity endpoints, and Lipinski rule-of-five compliance.

Comparative supervised classification workflow evaluating Random Forest, SVM, and CNNs on high-dimensional oncological expression matrices.

Community profiling pipeline computing alpha and beta diversity metrics (Shannon, Simpson, Bray-Curtis PCoA) for metagenomic datasets.

3D variant mapping and AlphaFold structural overlay evaluating residue perturbations and conformational stability changes.

Structure-based docking simulations mapping binding poses, affinities, and intermolecular contacts with therapeutic target pockets.

3D homology model construction with multiple sequence alignment and Ramachandran/ERRAT stereochemical validation.

Thermodynamic folding simulations calculating minimum free energy conformations, base-pairing probabilities, and mountain plots.