Search for a command to run...
Favourable docking poses and visually stable molecular dynamics trajectories are still widely used to advance cryptic-pocket binding hypotheses, yet in the cases examined here trajectory stability alone did not consistently distinguish supported from unsupported hypotheses because it reflects persistence of a simulated pose rather than whether an interaction survives stricter biophysical scrutiny. Here we present Gen2, a prespecified biophysical triage framework for evaluating ligand-pocket hypotheses in cryptic-pocket discovery. Rather than treating stable trajectories as sufficient evidence, Gen2 integrates contact persistence, replicate-aware interpretation, state conditioning, and local hydration readouts to determine whether a proposed interaction is supported, rejected, or left unresolved. We evaluated the framework across three evidential roles. In MYC Arg58Ala, where static site nomination, docking, and an initially plausible molecular-dynamics narrative suggested a putative cryptic-pocket interaction, Gen2 rejected the hypothesis and served as a negative-control challenge. In TP53 Y220C, a canonical mutant cavity system, Gen2 recovered the expected benchmark signal through state-conditioned hydration and contact persistence. In KRAS G12D, an anchor case with independent mechanistic support at the switch-II pocket locus, Gen2 recovered a physically interpretable local hydration signal under matched apo and holo conditions. Together, these cases support treating cryptic-pocket evaluation as a biophysical triage problem rather than relying on pose scoring or trajectory stability alone, and provide a more conservative basis for prioritising binding hypotheses before experimental follow-up. These conclusions are limited to the three evidential settings examined here and do not establish universal transfer across crypticpocket systems or substitute for direct biochemical or biophysical validation.