Resources chevron_right Protocols chevron_right MMseqs2
Protocol // Protein

MMseq2

Cluster proteins by sequence similarity at high speed and sensitivity using Many-against-Many sequence searching.

Protein codeGitHub languageWiki menu_bookPaper scheduleApril 3, 2026

Installation

conda install -c conda-forge -c bioconda mmseqs2

Cluster Proteins

Create database

This database should contain the proteins you want to cluster:

mmseqs createdb sequences.faa sequences_DB

Run the clustering

mmseqs cluster sequences_DB sequences_clu tmp --min-seq-id 0.3 -e 0.0001 --cluster-mode 2 --threads 20
info

In this example the minimum sequence identity is 30% and matches must have an e-value lower than 1E-05.

Generate a TSV file from the clusters

mmseqs createtsv sequences_DB sequences_DB sequences_clu sequences_clu.tsv

Generate other outputs

mmseqs createseqfiledb sequences_DB sequences_clu sequences_clu_seq
mmseqs result2flat sequences_DB sequences_DB sequences_clu_seq sequences_clu_seq.fasta