Modern AI-Assisted In-Silico Drug Discovery Workflows

Beginner-accessible, but not basic. This cohort starts with the workflow foundations and progressively moves toward modern validation, MD interpretation, affinity thinking, and decision-making.

What this cohort is really about

This is not a workshop where participants simply copy a docking command, run a short MD trajectory, generate RMSD/RMSF plots, and call the result a conclusion. The goal is to teach a stronger way to think about computational drug discovery: define the scientific question, understand the structure and biological context, choose tools deliberately, validate outputs, interpret uncertainty, and decide what claim is justified.

Participants will learn how classical tools and modern AI-assisted tools can work together. Docking, scoring, molecular dynamics, structure prediction, and affinity estimation are treated as parts of an evidence chain, not as independent magic answers.

Session 0 — Optional Beginner Primer

• What molecular docking actually predicts and what it does not predict.
• What molecular dynamics adds, and why simulation is not automatically proof.
• What docking scores mean, why they can mislead, and how beginners overinterpret them.
• The basic protein-ligand workflow map from target selection to interpretation.
• Common beginner mistakes: wrong protein state, incorrect ligand form, missing cofactors, blind water removal, and unvalidated poses.

Session 1 — Scientific Question & Structure Intelligence

• How to define the biological and computational question before selecting tools.
• PDB structures versus predicted structures: when each is useful and when each is risky.
• UniProt, PDB, SIFTS, residue numbering, isoforms, mutations, missing segments, and literature context.
• How to decide whether waters, cofactors, metals, ions, ligands, and alternate chains should be kept or removed.
• How to report uncertainty instead of hiding it behind a clean-looking structure.

Session 2 — Protein Curation, Ligand Preparation & Docking

• PDBFixer and MODELLER-style repair logic: when to repair, when to model, and when not to invent missing biology.
• Ligand preparation: stereochemistry, protonation, tautomer states, conformers, charges, and file format traps.
• Classical docking as a baseline hypothesis generator, not a final answer.
• Multi-engine docking and AI-assisted scoring: how to compare outputs without pretending agreement is proof.
• How to think about poses as hypotheses that need validation.

Session 3 — Pose Validation, MD Design & Interpretation

• Pose validation before MD: physical plausibility, chemical sanity, interactions, ligand strain, clashes, and geometry.
• PoseBusters-style checks, ProLIF interaction fingerprints, and visual inspection as complementary signals.
• MD as a computational experiment: why “100 ns” alone does not prove binding.
• Interaction persistence, ligand stability, pocket behavior, replicas, convergence, and failure modes.
• Affinity thinking: MM/PBSA carefully, AI affinity signals, OpenFE concepts, uncertainty, and decision logic.

Bonus Advanced Preview

The cohort also previews the scientific challenges that appear in advanced systems: membrane proteins, GPCRs, cofactors, ions, water networks, and complex biological environments. These topics show why biology resists shortcuts and why workflow design matters.

Tools & Materials Delivered

A practical participant kit, not just lectures.

Participants receive the practical workflow materials used during the cohort.

• Setup guides
• Example datasets
• Reproducible notebooks
• Validation checklists
• Pose-review templates
• MD interpretation templates
• Decision-report templates
• Certificate of completion

Who should join?

This cohort is designed for Master’s students, PhD researchers, postdocs, early-career scientists, academic labs, and professional teams working around pharmacy, biotechnology, medicinal chemistry, bioinformatics, computational biology, and drug discovery.

You do not need to be advanced. You do need to be willing to think critically about tools, assumptions, validation, and conclusions.