Clustering the protein universe of life using DIAMOND DeepClust

Benjamin J. Buchfink; Émile Barbé; Haim Ashkenazy; Klaus Reuter; John A. Kennedy; Hajk Georg Drost

doi:10.1038/s41592-026-03030-z

Clustering the protein universe of life using DIAMOND DeepClust

Benjamin J. Buchfink
, Émile Barbé
, Haim Ashkenazy
, Klaus Reuter
, John A. Kennedy
, Hajk Georg Drost (Lead / Corresponding author)

Computational Biology

Research output: Contribution to journal › Article › peer-review

1 Citation (Scopus)

8 Downloads (Pure)

Abstract

Relating billions of proteins across the tree of life remains a challenging task for comparative biosphere genomics and artificial intelligence-driven structure prediction. Here we present DIAMOND DeepClust, a cascaded, ultra-fast clustering method enabling planetary-scale organization of protein space, scaling to trillions of sequences while retaining sensitivity at low identity. Aggregating 19 billion biosphere proteins into 544 million nonsingleton clusters, we show that using our DeepClust database, available for download, can enhance structure prediction with AlphaFold2.

Original language	English
Pages (from-to)	724-727
Number of pages	4
Journal	Nature Methods
Volume	23
Issue number	4
Early online date	24 Mar 2026
DOIs	https://doi.org/10.1038/s41592-026-03030-z
Publication status	Published - Apr 2026

ASJC Scopus subject areas

Biotechnology
Biochemistry
Molecular Biology
Cell Biology

Access to Document

10.1038/s41592-026-03030-zLicence: CC BY

Final Published VersionFinal published version, 2.59 MBLicence: CC BY

Cite this

@article{9381d611b59b4ab882829ea5f1667763,

title = "Clustering the protein universe of life using DIAMOND DeepClust",

abstract = "Relating billions of proteins across the tree of life remains a challenging task for comparative biosphere genomics and artificial intelligence-driven structure prediction. Here we present DIAMOND DeepClust, a cascaded, ultra-fast clustering method enabling planetary-scale organization of protein space, scaling to trillions of sequences while retaining sensitivity at low identity. Aggregating 19 billion biosphere proteins into 544 million nonsingleton clusters, we show that using our DeepClust database, available for download, can enhance structure prediction with AlphaFold2.",

author = "Buchfink, \{Benjamin J.\} and {\'E}mile Barb{\'e} and Haim Ashkenazy and Klaus Reuter and Kennedy, \{John A.\} and Drost, \{Hajk Georg\}",

note = "Copyright: {\textcopyright} The Author(s) 2026.",

year = "2026",

month = apr,

doi = "10.1038/s41592-026-03030-z",

language = "English",

volume = "23",

pages = "724--727",

journal = "Nature Methods",

issn = "1548-7091",

publisher = "Nature Research",

number = "4",

}

TY - JOUR

T1 - Clustering the protein universe of life using DIAMOND DeepClust

AU - Buchfink, Benjamin J.

AU - Barbé, Émile

AU - Ashkenazy, Haim

AU - Reuter, Klaus

AU - Kennedy, John A.

AU - Drost, Hajk Georg

PY - 2026/4

Y1 - 2026/4

N2 - Relating billions of proteins across the tree of life remains a challenging task for comparative biosphere genomics and artificial intelligence-driven structure prediction. Here we present DIAMOND DeepClust, a cascaded, ultra-fast clustering method enabling planetary-scale organization of protein space, scaling to trillions of sequences while retaining sensitivity at low identity. Aggregating 19 billion biosphere proteins into 544 million nonsingleton clusters, we show that using our DeepClust database, available for download, can enhance structure prediction with AlphaFold2.

AB - Relating billions of proteins across the tree of life remains a challenging task for comparative biosphere genomics and artificial intelligence-driven structure prediction. Here we present DIAMOND DeepClust, a cascaded, ultra-fast clustering method enabling planetary-scale organization of protein space, scaling to trillions of sequences while retaining sensitivity at low identity. Aggregating 19 billion biosphere proteins into 544 million nonsingleton clusters, we show that using our DeepClust database, available for download, can enhance structure prediction with AlphaFold2.

UR - https://www.scopus.com/pages/publications/105034322645

U2 - 10.1038/s41592-026-03030-z

DO - 10.1038/s41592-026-03030-z

M3 - Article

C2 - 41876643

AN - SCOPUS:105034322645

SN - 1548-7091

VL - 23

SP - 724

EP - 727

JO - Nature Methods

JF - Nature Methods

IS - 4

ER -

Clustering the protein universe of life using DIAMOND DeepClust

Abstract

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this