#40 · Structural bio & protein design
Sampling strategy governs which conformational states biomolecular methods recover
The dominant thread is that the sampling procedure, more than the underlying model alone, determines which biomolecular conformations are recovered. In a T4 lysozyme benchmark, generative methods mostly missed buried features when sampling conformational states, as compared against AI-accelerated simulation and enhanced MD. Ensembles from AlphaFold3, AlphaFold2, and RosettaFold were scored against NMR, X-ray, cryo-EM, and MD flexibility profiles, with AF3 performing best. For intrinsically disordered proteins of 15 to 71 residues, OPES multithermal enhanced sampling in a single replica produces atomistic ensembles that reveal transient tertiary structure. What is new is explicit attention to the inference layer as a determinant of results across these studies. The evidence remains preprint-dominated, with only the OPES study published.
Published papers
Transient tertiary structure in intrinsically disordered proteins revealed by multithermal enhanced sampling.
OPES multithermal enhanced sampling in one replica yields atomistic ensembles for disordered proteins (15-71 residues), revealing transient tertiary structure…
Preprints
Benchmarking generative AI and physics based molecular simulation for sampling conformational heterogeneity in T4 Lysozyme
Benchmarks generative AI, AI-accelerated simulation, and enhanced MD for sampling T4 lysozyme conformational states; generative methods mostly miss buried…
Deep Learning Structural Ensembles as Proxies for Protein Flexibility
Benchmarks AlphaFold3, AlphaFold2 and RosettaFold ensembles against NMR, X-ray, cryo-EM and MD flexibility profiles, finding AF3 best; provides Jupyter…
Topic history
| Date | Rank |
|---|---|
| #40 | |
| #24 | |
| #19 | |
| #9 | |
| #22 | |
| #12 | |
| #11 | |
| #6 | |
| #31 | |
| #27 | |
| #19 | |
| #39 | |
| #37 | |
| #39 | |
| #27 | |
| #20 | |
| #21 | |
| #30 | |
| #19 | |
| #9 | |
| #15 | |
| #15 | |
| #16 | |
| #9 | |
| #15 | |
| #19 | |
| #14 | |
| #27 | |
| #19 | |
| #26 | |
| #23 | |
| #31 | |
| #24 | |
| #27 | |
| #33 |