The abstract is read at the publisher; the summary is JClub's.
Lacuna’s conformational-ensemble pocket discovery recovers cryptic binding sites in CryptoBench’s designated test fold for 59.9% of 222 structures in top five predictions using the zero-dependency default.
- Why it matters: Static predictors miss cryptic pockets that are absent or too small in the unbound structure, so ensemble-aware detection is needed to expose transient binding sites.
- What they did: Lacuna builds a conformational ensemble from any input (default NMA), detects pockets in every conformer, clusters persistent sites, and ranks candidates with a fitted model using within-structure pairs.
- The result: On that fold, replacing the prior analytic ordering yields a +27.5 point gain (95% CI +21.2 to +33.8), and the ensemble improves per-residue propensity from AUROC 0.249 to 0.766 (AUPRC 0.340).
The findingWhy it mattersWhat they didThe result
AI summary of the abstract · may contain errors · not medical advice · read the paper.