Sampling only side chains of an apo structure recovers many more docking poses than apo or MD ensembles
- A Monte Carlo protocol perturbs pocket side-chain torsions, interleaved with short restrained MD relaxation, to build receptor ensembles from an unbound structure.
- Ensembles were tested retrospectively on several hundred protease complexes against the apo receptor and restrained or unrestrained MD.
- Correct poses were recovered for clearly more complexes, especially with large side-chain motion, while pockets stayed realistic. Gains reflect best-case pose coverage; ranking and screening remain untested.
💡 Points Based on the abstract
Does better pose recovery mean better screening? Not yet. Coverage counts a success if any ensemble member docks correctly, so bigger ensembles get more shots. Without a size-matched control or decoy enrichment the gain is hard to price. Next, dock known actives and property-matched decoys into these ensembles and see whether enrichment improves or false positives grow.