Program Structure

Module 1: Drug Discovery Basics + Where CADD Fits

• Drug discovery pipeline overview: target → hit → lead → optimization.
• What CADD can accelerate (and what it cannot replace).
• Key concepts: binding affinity, selectivity, drug-likeness, ADMET thinking.
• Understanding datasets and confidence: crystal structures vs predicted models.

Module 2: Protein & Ligand Preparation (The Step Most People Get Wrong)

• Fetching protein structures (PDB) and understanding chain/active-site selection.
• Cleaning proteins: water removal decisions, missing residues, protonation, charges.
• Ligand preparation: SMILES → 3D, tautomer/ionization states, energy minimization.
• Defining binding sites: co-crystal ligands, pocket prediction, grid box setup.

Module 3: Molecular Docking Workflow

• Docking fundamentals: scoring functions, search algorithms, pose generation.
• Running docking experiments (single ligand and batch docking).
• Pose selection: docking score vs interaction quality (how to avoid false confidence).
• Interaction analysis: H-bonds, hydrophobic contacts, salt bridges, π-π stacking.

Module 4: Docking Validation & Post-Docking Filtering

• Redocking and RMSD: checking if your docking is reliable.
• Consensus docking (concept) and rescoring basics.
• Filtering hits: Lipinski/Veber rules, PAINS alerts (concept), basic toxicity flags.
• Prioritizing compounds for wet-lab testing: practical decision framework.

Module 5: QSAR Fundamentals (From Chemistry to Predictive Models)

• What QSAR predicts and what it needs: curated data, consistent endpoints.
• Data preprocessing: duplicates, outliers, activity transformations (pIC50).
• Descriptors and fingerprints: 1D/2D/3D descriptors, Morgan fingerprints.
• Train/validation/test split strategies and why random split can fail.

Module 6: Building QSAR Models (Hands-on ML)

• Baseline models: Linear regression, Random Forest, SVR (concept + practice).
• Classification vs regression QSAR: when to use which.
• Model evaluation: R², RMSE, MAE, ROC-AUC (and what “good” really means).
• Feature importance and interpretability: linking chemistry to predictions.

Module 7: Pharmacophore Modeling & Virtual Screening

• Pharmacophore basics: H-bond donors/acceptors, hydrophobics, aromatic features.
• Ligand-based pharmacophore: building from active compound sets.
• Structure-based pharmacophore: using protein–ligand interactions.
• Virtual screening workflow: pharmacophore filter → docking → rescoring.

Module 8: Mini-Pipeline Integration (How Real CADD is Done)

• Designing a combined workflow: pharmacophore + docking + QSAR prioritization.
• Hit ranking and shortlisting with multi-criteria decision-making.
• Reproducible reporting: parameters, versions, and result traceability.
• Writing a discovery-style report: methods, results, limitations, next steps.

Final Project

• Select a target (enzyme/receptor) and a compound set (known actives + candidates).
• Perform docking and interaction analysis to shortlist hits.
• Build a QSAR model (or pharmacophore model) to prioritize candidates.
• Deliverables: full workflow report + figures/tables + shortlist rationale.
• Example projects: docking inhibitors for a cancer target, QSAR for antimicrobial activity, pharmacophore for kinase inhibitors.