Several multiple sequence alignment perturbation methods enhance AlphaFold3 sampling of alternative protein states

2026 · doi:10.64898/2026.04.02.716037
preprint OA: closed
Full text JSON View at publisher
⚙ AI-generated deep summary by claude@2026-06, 2026-06-24 · read from full text ⓘ

The study evaluates whether multiple sequence alignment (MSA) perturbation strategies—such as stochastic subsampling, clustering, and column masking—enhance AlphaFold3 (AF3) sampling of alternative protein conformational states, compared with AlphaFold2 (AF2) and the BioEmu Boltzmann sampling model. Using 107 proteins with multiple experimentally solved conformations, the authors report that unperturbed AF3 already samples alternative states with higher TM-scores than AF2 and roughly comparable performance to BioEmu, and that applying any MSA perturbation improves AF3 sampling in a statistically significant way in roughly 20% of cases, with improvements in the top 1% TM-score of at least 0.05 and rare worsening. A limitation is that the evaluation is based on these curated protein sets and uses TM-score–based comparisons rather than broader functional or cell-context validation. This paper does not explicitly discuss endometriosis or adenomyosis; it was included in the corpus via a keyword match in the upstream search index.

Read from the paper's body, not the abstract. Not a substitute for reading the paper. No clinical advice. How this works

Abstract

Protein function often involves multiple conformational states. Several multiple sequence alignment-perturbing strategies, including stochastic subsampling, clustering, and column masking, have been shown to enhance AlphaFold2 (AF2) sampling of alternative protein states. Here, we evaluate these strategies on AlphaFold3 (AF3) and compare their performance with the BioEmu Boltzmann sampling model on 107 proteins with multiple experimentally solved conformational states. We find that unperturbed AF3 samples alternative states with significantly higher TM-scores compared to AF2 and comparable to BioEmu. In particular, all MSA perturbation methods improve AF3 sampling at a statistically significant level, improving the top 1% TM-score by at least 0.05 in approximately 20% of cases each, while rarely worsening the performance. Furthermore, we find that different choices of amino acid masks can improve column-masked AF3 sampling for specific targets. Our results highlight how MSA perturbations remain relevant in AF3, providing a useful tool for understanding dynamic biological processes.
Full text 1,292 characters · extracted from oa-doi-fallback · click to expand
Abstract Protein function often involves multiple conformational states. Several multiple sequence alignment-perturbing strategies, including stochastic subsampling, clustering, and column masking, have been shown to enhance AlphaFold2 (AF2) sampling of alternative protein states. Here, we evaluate these strategies on AlphaFold3 (AF3) and compare their performance with the BioEmu Boltzmann sampling model on 107 proteins with multiple experimentally solved conformational states. We find that unperturbed AF3 samples alternative states with significantly higher TM-scores compared to AF2 and comparable to BioEmu. In particular, all MSA perturbation methods improve AF3 sampling at a statistically significant level, improving the top 1% TM-score by at least 0.05 in approximately 20% of cases each, while rarely worsening the performance. Furthermore, we find that different choices of amino acid masks can improve column-masked AF3 sampling for specific targets. Our results highlight how MSA perturbations remain relevant in AF3, providing a useful tool for understanding dynamic biological processes. - AlphaFold - MSA perturbations - protein structure prediction - alternative state - protein conformations Competing Interest Statement The authors have declared no competing interest.

Text is read by the "Ask this paper" AI Q&A widget below. Extraction quality varies by source — PMC NXML preserves structure cleanly, OA-HTML may include some navigation residue, and OA-PDF can have broken hyphenation. The publisher copy (via DOI) is the canonical version.

My notes (saved in your browser only)

⚙ Ask this paper AI returns verbatim quotes from the full text · source: oa-doi-fallback ⓘ

Answers must be backed by verbatim quotes from this paper's full text. Hallucinated quotes are dropped automatically; if no verbatim passage answers the question, we say so. How this works

Citation neighborhood (no data yet)

We don't have any in-corpus citations linked to this paper yet. This is a recent paper (2026) — citers typically take a year or two to land, and the OpenAlex reference graph may still be filling in.

Source provenance

europepmc
last seen: 2026-05-20T01:45:00.602351+00:00