EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

ProtMamba: a homology-aware but alignment-free protein state space model

Damiano Sgarbossa iD, Cyril Malbranke iD, Anne-Florence Bitbol iD

Bioinformatics · 02.06.2025 · 41 (6)

DOI: 10.1093/bioinformatics/btaf348 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Bioinformatics, online erschienen am 13.06.2025. Verfasst von Damiano Sgarbossa und 2 weiteren Autor:innen. Beteiligte Einrichtung: Institute of Bioengineering, School of Life Sciences, École Polytechnique Fédérale de Lausanne (EPFL), CH-1015 Lausanne, Switzerland. Hauptthemen laut MEDLINE: Proteins, Software, Sequence Alignment, Sequence Analysis, Protein. Die Arbeit wurde bisher 21-mal zitiert (Crossref) und nennt 56 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: European Research Council, European Union's Horizon 2020 research and innovation programme, European Union’s Horizon 2020 research and innovation programme.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Motivation Protein language models are enabling advances in elucidating the sequence-to-function mapping, and have important applications in protein design. Models based on multiple sequence alignments efficiently capture the evolutionary information in homologous protein sequences, but multiple sequence alignment construction is imperfect. Results We present ProtMamba, a homology-aware but alignment-free protein language model based on the Mamba architecture. In contrast with attention-based models, ProtMamba efficiently handles very long context, comprising hundreds of protein sequences. It is also computationally efficient. We train ProtMamba on a large dataset of concatenated homologous sequences, using two GPUs. We combine autoregressive modeling and masked language modeling through a fill-in-the-middle training objective. This makes the model adapted to various protein design applications. We demonstrate ProtMamba’s usefulness for sequence generation, motif inpainting, fitness prediction, and modeling intrinsically disordered regions. For homolog-conditioned sequence generation, ProtMamba outperforms state-of-the-art models. ProtMamba’s competitive performance, despite its relatively small size, sheds light on the importance of long-context conditioning. Availability and implementation A Python implementation of ProtMamba is freely available in our GitHub repository: https://github.com/Bitbol-Lab/ProtMamba-ssm and archived at https://doi.org/10.5281/zenodo.15584634.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

ProteinsSoftwareSequence AlignmentSequence Analysis, ProteinComputational BiologyAlgorithms

Substanzen

Proteins

Autor:innen und Institutionen

  • Institute of Bioengineering, School of Life Sciences, École Polytechnique Fédérale de Lausanne (EPFL), CH-1015 Lausanne, Switzerland.
  • SIB Swiss Institute of Bioinformatics, CH-1015 Lausanne, Switzerland.

ORCID-Profile: Damiano Sgarbossa, Cyril Malbranke, Anne-Florence Bitbol

Förderung

  • European Research Council
  • European Union's Horizon 2020 research and innovation programme
  • European Union’s Horizon 2020 research and innovation programme

Zitieren

APA

Sgarbossa, D., Malbranke, C., & Bitbol, A. F. (2025). ProtMamba: a homology-aware but alignment-free protein state space model. Bioinformatics, 41(6). https://doi.org/10.1093/bioinformatics/btaf348

Vancouver

Sgarbossa D, Malbranke C, Bitbol AF. ProtMamba: a homology-aware but alignment-free protein state space model. Bioinformatics. 2025;41(6). doi:10.1093/bioinformatics/btaf348

BibTeX
@article{Sgarbossa2025,
  author = {Sgarbossa, Damiano and Malbranke, Cyril and Bitbol, Anne-Florence},
  title = {ProtMamba: a homology-aware but alignment-free protein state space model},
  journal = {Bioinformatics},
  year = {2025},
  volume = {41},
  number = {6},
  doi = {10.1093/bioinformatics/btaf348}
}
Kostenlos starten →