Questions: Ligand Binding and Docking

Question 1

A virtual screening campaign docks 1 million compounds and selects the top 100 by docking score. Approximately what hit rate (fraction of experimentally confirmed binders) would you expect?

• A. Nearly 100% — docking accurately predicts all binders
• B. Approximately 1-10% — docking enriches for potential binders compared to random selection but has a high false-positive rate due to scoring function limitations
• C. 0% — docking never identifies real binders
• D. 50% — docking is equivalent to random selection

Answer: (B)

Docking-based virtual screening typically achieves hit rates of 1-10%, compared to ~0.01-0.1% for random screening — an enrichment of 10-1000x. This enrichment is valuable (it dramatically reduces the number of compounds that need experimental testing) but the high false-positive rate reflects the limitations of scoring functions: they estimate binding energy from approximate terms and miss contributions from protein flexibility, water networks, entropy, and strain. Post-docking filtering (pharmacokinetic properties, visual inspection of binding poses, consensus scoring with multiple methods) can improve hit rates. Docking is a hypothesis-generating tool that narrows the search space, not a quantitative affinity predictor.

Question 2

Molecular docking can accurately predict the binding free energy (Kd) of a ligand from the docking score alone.

• T. True
• F. False

Answer: False

Docking scoring functions (empirical, force-field-based, or knowledge-based) are designed to distinguish binders from non-binders and to predict binding poses, but they are poor at quantitatively predicting binding free energies. The correlation between docking scores and experimental binding affinities is typically weak (R^2 ~ 0.1-0.3). This is because scoring functions use simplified energy terms, do not account for protein conformational changes upon binding (induced fit), poorly estimate the entropic cost of binding, and inadequately model water-mediated interactions. More rigorous methods (FEP, thermodynamic integration) provide better affinity predictions but at computational costs 10,000-100,000x higher than docking.

Question 3

Explain the concept of induced fit and why it is a challenge for molecular docking.

Model answer: Induced fit describes the conformational change that occurs in the protein upon ligand binding — the binding site reshapes itself to accommodate the ligand, forming interactions that are not present in the unbound (apo) structure. Standard docking treats the protein as rigid (fixed atomic coordinates), docking the ligand into the apo binding site. If the true binding mode requires a protein conformational change (loop rearrangement, side chain rotation, domain closure), rigid docking will either miss the correct pose or score it poorly because it cannot account for the protein's adaptation. Solutions include flexible receptor docking (allowing specified side chains to move), ensemble docking (docking into multiple protein conformations), and induced-fit docking protocols (iterating between ligand docking and protein conformational adjustment).

Many important drug-target interactions involve induced fit: kinase inhibitors that bind the DFG-out conformation, HIV protease inhibitors that cause flap closure, and nuclear hormone receptor agonists that reposition the activation helix. For these targets, docking into a single crystal structure systematically misses binding modes that require protein conformational change.