Ensemble tests mask missing dynamics in protein conformational generators
- 1Zhejiang University
Abstract
Protein generators are increasingly used to produce conformational ensembles and trajectories as faster alternatives to molecular dynamics (MD). Yet whether agreement with an equilibrium ensemble demonstrates that a generator has learned dynamics remains unresolved. Here we show that this evidence survives the destruction of dynamics. Randomly reordering authentic MD frames leaves ensemble fidelity unchanged but reduces the geometrically gated kinetic pass rate from 0.963 to 0.004; an independent MD replicate reaches 0.95 under the same test. Dynbench separates recovery of conformational states, time-dependent behaviour and trajectory admissibility. Public generators that appear similar under ensemble tests separate sharply, and in a prospective 36-protein lockbox the reference and floor remain stable while two models reverse order. Ensemble tests can therefore systematically overstate evidence of learned dynamics, establishing time-resolved validation as a distinct requirement for claims about protein dynamics generators.
Related
- TopoFlow: Evolutionarily Conditioned Flow Matching for Protein Conformational Ensemble Generation
- trFlow: ultrafast all-atom protein conformational ensemble generation via conditional flow matching
- Kryptix-1: Conformational Ensemble Sampling Substantially Improves Cryptic Pocket Detection Relative to Static-Structure and Prior Computational Approaches
- Lacuna: Cryptic Binding Pocket Discovery via Conformational Ensemble Analysis
- Targeted finetuning enables co-folding models to learn ligand-induced protein conformational states
BibTeX
@misc{liu2026ensemble,
title = {{Ensemble tests mask missing dynamics in protein conformational generators}},
author = {K. Liu and Q. Qian and Y. Chi},
howpublished = {bioRxiv},
year = {2026},
month = sep,
doi = {10.64898/2026.09.02.748111},
url = {https://www.biorxiv.org/content/10.64898/2026.09.02.748111v1},
note = {CC BY},
}
RIS
TY - UNPB
TI - Ensemble tests mask missing dynamics in protein conformational generators
AU - K. Liu
AU - Q. Qian
AU - Y. Chi
PY - 2026
DA - 2026/09/02
PB - bioRxiv
DO - 10.64898/2026.09.02.748111
UR - https://www.biorxiv.org/content/10.64898/2026.09.02.748111v1
AB - Protein generators are increasingly used to produce conformational ensembles and trajectories as faster alternatives to molecular dynamics (MD). Yet whether agreement with an equilibrium ensemble demonstrates that a generator has learned dynamics remains unresolved. Here we show that this evidence survives the destruction of dynamics. Randomly reordering authentic MD frames leaves ensemble fidelity unchanged but reduces the geometrically gated kinetic pass rate from 0.963 to 0.004; an independent MD replicate reaches 0.95 under the same test. Dynbench separates recovery of conformational states, time-dependent behaviour and trajectory admissibility. Public generators that appear similar under ensemble tests separate sharply, and in a prospective 36-protein lockbox the reference and floor remain stable while two models reverse order. Ensemble tests can therefore systematically overstate evidence of learned dynamics, establishing time-resolved validation as a distinct requirement for claims about protein dynamics generators.
ER -