Do AF2 decoys correspond to metastable basins and include required metastable states for drug design?

Ascertain whether decoy structures generated by reduced MSA AlphaFold2 ensembles correspond to physically meaningful metastable basins and whether these decoys include structures representing the specific metastable states required for the intended drug design targets.

Background

Within the context of kinase conformational heterogeneity, the authors use rMSA AF2 to produce large decoy ensembles and investigate whether such decoys map onto actual metastable free-energy basins. They further question whether these decoys contain structures that match the metastable conformations required for specific ligand classes (e.g., those targeting inactive states).

Clarifying this correspondence is essential for deciding if AF2-derived decoys can be used in ensemble docking and for prioritizing structures that represent therapeutically relevant metastable states.