📚 Syllabus & Course Curriculum
AI & Machine Learning in Healthcare
Module-by-module breakdown of AI-Enabled CADD & Machine Learning for Drug Design, from foundations to a certified capstone project.
Targets
• Binding site detection with fpocket and the risk of screening a crystallographic artefact
• Protonation states, tautomers and missing loops as the usual source of bad docking
• Holo versus apo structures and why induced fit breaks rigid-receptor assumptions
Libraries
• ZINC and ChEMBL as sources, and the difference between purchasable and virtual
• Lipinski, Veber and PAINS filters — what each removes and what each wrongly removes
• 3D conformer generation with RDKit and the conformer-count trade-off
Docking
• AutoDock Vina and Glide scoring functions and their weak correlation with affinity
• Redocking and cross-docking as the only honest validation of a docking protocol
• Enrichment metrics: ROC AUC and BEDROC against DUD-E style decoys
Modelling
• Molecular descriptors and fingerprints (ECFP) versus learned graph representations
• Scaffold splits rather than random splits, or the model reports fantasy accuracy
• ADMET endpoints — hERG, CYP inhibition, solubility — and the applicability domain
Practice
• Hit triage: consensus scoring, visual inspection and chemical common sense
• MM-GBSA rescoring and short MD to test pose stability before committing
• Documenting a screen so the result can be reproduced by someone else
Earn government-registered certification in AI-Enabled CADD & Machine Learning for Drug Design
e-Certificate and e-Marksheet issued on successful completion.
