Alternate conformations can be sampled with MSA-based structure prediction methods using custom PDB databases and subsampled MSAs

Summary

Alternate conformations can be sampled with MSA-based methods AlphaFold2 and Boltz by using either custom templates (1) or custom sequences databases (2) or MSAs (3,1). For the last case, MSAs are modified either by clustering using HDBSCAN or randomly subsampling, respectively. However, these ensembles do not correspond to the energetics of those proteins. MSA-based tricks were recently shown to work with Boltz (4). These hacks circumvent the broader tendency of structure prediction methods to undersample conformations they find to be high-confidence.

Details

Alternate conformations sampled via

• Subsampled MSAs: Fewer sequences are passed in the MSA (1). Required for addition of other restraints (e.g., via AlphaLink, (5)). Also works for CDRs of nanobodies (6).
• Masked MSA columns: Entire columns of residues are mutated to either alanine or masked (7,8)
• Clustered MSAs: (3)
• Template curation: Can more reliably influence which conformation is sampled, but requires an experimental structure (though Faezov and (9) used models for some of their kinases).
• Dropout: Shown by (8) to not be very effective, unlike for docking (link)

(10) Vani et al summarized the challenge of MSA subsampling as follows:

“The structures obtained, including those that are metastable, are not in any physically reasonable probability distribution. Nor is there an obvious way to directly obtain a distribution or free energy surface from them that could account for both enthalpy and entropy.”

Figures

Ref (6)

Ref (11)

1.

del Alamo D, Sala D, Mchaourab HS, Meiler J. Sampling alternative conformational states of transporters and receptors with AlphaFold2. eLife. 2022;11. Available from: https://doi.org/10.7554/elife.75751

2.

Monteiro da Silva G, Cui JY, Dalgarno DC, Lisi GP, Rubenstein BM. High-throughput prediction of protein conformational distributions with subsampled AlphaFold2. Nature Communications. 2024;15(1). Available from: https://doi.org/10.1038/s41467-024-46715-9

3.

Wayment-Steele HK, Ojoawo A, Otten R, Apitz JM, Pitsawong W, Hömberger M, et al. Predicting multiple conformations via sequence clustering and AlphaFold2. Nature. 2023;625(7996):832–9. Available from: https://doi.org/10.1038/s41586-023-06832-9

4.

Richman DD, Karaguesian J, Suomivuori C-M, Dror RO. Unlocking hidden biomolecular conformational landscapes in diffusion models at inference time. In: NeurIPS 2025. 2025. Available from: https://openreview.net/forum?id=U87XyMPrZp

5.

Stahl K, Graziadei A, Dau T, Brock O, Rappsilber J. Protein structure prediction with in-cell photo-crosslinking mass spectrometry and deep learning. Nature Biotechnology. 2023;41(12):1810–9. Available from: https://doi.org/10.1038/s41587-023-01704-z

6.

Riccabona JR, Spoendlin FC, Fischer A-LM, Loeffler JR, Quoika PK, Jenkins TP, et al. Assessing AF2’s ability to predict structural ensembles of proteins. Structure. 2024;32(11):2147-2159.e2. Available from: https://doi.org/10.1016/j.str.2024.09.001

7.

Stein RA, Mchaourab HS. SPEACH_AF: Sampling protein ensembles and conformational heterogeneity with Alphafold2. PLOS Computational Biology. 2022;18(8):e1010483. Available from: https://doi.org/10.1371/journal.pcbi.1010483

8.

Kalakoti Y, Wallner B. AFsample2 predicts multiple conformations and ensembles with AlphaFold2. Communications Biology. 2025;8(1). Available from: https://doi.org/10.1038/s42003-025-07791-9

9.

Faezov B, Dunbrack RL. AlphaFold2 models of the active form of all 437 catalytically competent human protein kinase domains. openRxiv; 2023. Available from: https://doi.org/10.1101/2023.07.21.550125

10.

Vani BP, Aranganathan A, Tiwary P. Exploring Kinase Asp-Phe-Gly (DFG) Loop Conformational Stability with AlphaFold2-RAVE. Journal of Chemical Information and Modeling. 2023;64(7):2789–97. Available from: https://doi.org/10.1021/acs.jcim.3c01436

11.

Schafer JW, Porter LL. AlphaFold2’s training set powers its predictions of fold-switched conformations. openRxiv; 2024. Available from: https://doi.org/10.1101/2024.10.11.617857