GPU-accelerated homology search with MMseqs2
Rapidly growing protein databases demand faster sensitive search tools. Here the graphics processing unit (GPU)-accelerated MMseqs2 delivers 6× faster single-protein searches than CPU methods on 2 × 64 cores, speeds previously requiring large protein batches. For larger query batches, it is the most cost-effective solution, outperforming the fastest alternative method by 2.4-fold with eight GPUs. It accelerates protein structure prediction with ColabFold 31.8× over the standard AlphaFold2 pipeline and protein structure search with Foldseek by 4-27×. MMseqs2-GPU is available under an open-source license at https://mmseqs.com/ .
Authors
- Milot Mirdita (ORCID: https://orcid.org/0000-0001-8637-6719)
- Martin Steinegger (ORCID: https://orcid.org/0000-0001-8781-9753)
- Bertil Schmidt (ORCID: https://orcid.org/0000-0003-2597-8331)
- Christian Hundt (ORCID: https://orcid.org/0000-0002-7995-1858)
- Kieran Didi (ORCID: https://orcid.org/0000-0001-6839-3320)
- Sooyoung Cha (ORCID: https://orcid.org/0000-0001-7211-4603)
- Alejandro Chacón (ORCID: https://orcid.org/0000-0001-8851-7618)
- Felix Kallenborn (ORCID: https://orcid.org/0000-0003-4516-6357)
- Christian Dallago (ORCID: https://orcid.org/0000-0003-4650-6181)
- Hassan Sirelkhatim (ORCID: https://orcid.org/0009-0005-0944-3005)
Institutions
- Seoul National University (KR)
- Duke University (US)
- Johannes Gutenberg University Mainz (DE)
- University of Oxford (GB)
- National University College (PR)
- Nvidia (United States) (US)
- Institute of Molecular Biology (AM)
Publication Details
- Journal
- Nature Methods
- Published
- 2025-09-18
- DOI
- https://doi.org/10.1038/s41592-025-02819-8
- Citations
- 22
- Primary Topic
- Genomics and Phylogenetic Studies
- Type
- article
- Field-Weighted Citation Impact
- 17.73
Funders
- National Research Foundation
- Deutsche Forschungsgemeinschaft
- National Research Foundation of Korea
- Johannes Gutenberg-Universität Mainz
- Novo Nordisk
- Novo Nordisk Fonden
- Samsung