Benchmarking generative models of protein conformational ensembles against all-atom molecular dynamics.
| Model | Data | RMWD ↓ | RMSF r ↑ | Time/chain ↓ | N |
|---|---|---|---|---|---|
| AlphaFlow-MD | held-out (weak) | 3.81 | 0.814 | n/a | 82 |
| AlphaFlow-MD (distilled) | held-out (weak) | 4.76 | 0.783 | n/a | 82 |
| ESMFlow-MD | held-out (weak) | 5.98 | 0.704 | n/a | 82 |
| BioEmu | held-out | 6.17 | 0.751 | 718s | 81 |
| ESMFlow-MD (distilled) | held-out (weak) | 6.43 | 0.699 | 365s | 82 |
| ESMDiff | uncertain | 7.74 | 0.664 | 237s | 82 |
RMWD root-mean Wasserstein distance (Å), lower is better · RMSF r per-residue flexibility correlation, higher is better · Time/chain median inference time per chain (s), lower is faster · N chains scored
Data how strong is the ATLAS held-out guarantee? · held-out never trained on ATLAS/MD · held-out (weak) trained on ATLAS train, temporal split only · uncertain unknown
Browse chains →Scored on the ATLAS test split, against all-atom MD trajectories from the ATLAS dataset.