MMseqs2

Overview

Creating A3M alignments for structure predictors.
Searching sequence collections for homologs.
Clustering related protein sequences.
Preparing sequence sets for downstream design or analysis.

Modes

Mode	Input shape	When to use it
`msa` Build MSA Default	Uses one selected file/source when file parameters are present.	Generate a predictor-ready A3M multiple sequence alignment from one FASTA/FASTQ file.
`search` Search Database	Uses one selected file/source when file parameters are present.	Search sequences against a managed reference database and write a TSV hit table.
`cluster` Cluster Sequences	Uses one selected file/source when file parameters are present.	Cluster sequences from one FASTA/FASTQ file and write cluster outputs.
`enrich_subseq_seqs` Add MSAs to SubSeq Seqs	Consumes a folder or output set; useful for batches and pipeline handoffs.	Read SubSeq seq JSON files from a folder, compute shared MSAs for selected protein entities, and write enriched SubSeq seq outputs.

Canonical Job Configuration

These are the fields exposed by the default job configuration for mmseqs2. They are also returned by GET /api/v1/program/params?program=mmseqs2 and submitted as the params JSON object to POST /api/v1/job/submit.

Parameter	Type	Modes	What it does
`sequence_file` Sequence File	Sequence file	Build MSA, Search Database, Cluster Sequences	Paste a protein sequence or choose a FASTA/FASTQ file from an upload, dataset, or previous job output. Required; Files: .fasta, .fa, .faa, .fas, .fna, .ffn, .fastq, .fq
`reference_database` Reference Database	Text	Build MSA, Search Database, Add MSAs to SubSeq Seqs	Managed reference database used for MSA generation or sequence search. Default: UniRef30; Options: UniRef30, UniRef50, UniRef90, Swiss-Prot, PDB, Environmental DB
`search_goal` Search Goal	Text	Search Database	Use homology search for remote matches, or high-identity mapping for very similar sequences. Default: Homology search; Options: Homology search, High-identity mapping
`sensitivity` Sensitivity	Text	Build MSA, Search Database, Add MSAs to SubSeq Seqs	Higher sensitivity finds more distant homologs but costs more runtime. Default: Balanced; Options: Fast, Balanced, Sensitive; Shown when search_goal is Homology search
`output_detail` Output Detail	Text	Search Database	Choose a compact hit table or include aligned query and target residue strings. Default: Summary table; Options: Summary table, Include aligned residues
`clustering_scale` Clustering Scale	Text	Cluster Sequences	Standard clustering is more sensitive; huge-dataset clustering uses MMseqs2 linear-time clustering. Default: Standard; Options: Standard, Huge dataset
`minimum_identity` Minimum Identity	Number	Cluster Sequences	Minimum sequence identity for linking sequences into the same cluster. Default: 0.5; Range: 0-1
`minimum_coverage` Minimum Coverage	Number	Cluster Sequences	Minimum aligned coverage for clustering, using MMseqs2's bidirectional full-length overlap rule. Default: 0.8; Range: 0-1
`result_name` Result Name	Text	All modes	Optional output name. The job form writes the result with the appropriate extension or prefix.
`entity_ids` Entities To Enrich	Text	Add MSAs to SubSeq Seqs	Comma-separated entity IDs to receive precomputed MSAs in each .seq.json file. Use all to enrich every protein entity. Required
`empty_other_msas` Empty Other MSAs	Yes/no	Add MSAs to SubSeq Seqs	Set non-selected protein entities to no-MSA mode, useful for target:binder screening where only the target should use an MSA. Default: true
`reuse_existing_msas` Reuse Existing MSAs	Yes/no	Add MSAs to SubSeq Seqs	Keep existing precomputed MSA references instead of recomputing them for selected entities. Default: true

Advanced configuration fields

No additional fields in this group.

Outputs And Metrics

Search results, alignment files, cluster outputs, and converted FASTA/A3M-style files depending on mode.
Sequence-search outputs should be interpreted by e-value, identity, coverage, and alignment context.

Common Examples

Build an MSA for one protein FASTA, then use the A3M in Boltz-2 or AlphaFold-style workflows.
Cluster a FASTA library before selecting representatives.

Example API params

{
  "mode": "msa",
  "sequence_file": "query.fasta",
  "reference_database": "UniRef30",
  "sensitivity": "Balanced"
}

Caveats

Alignment quality depends on database choice, search sensitivity, and query composition.
Low-complexity and repeated regions can produce misleading hits.

Advanced Submit

Advanced submit is still available for direct program arguments through POST /api/v1/job/submit-advanced. Prefer canonical configuration unless you need exact low-level arguments or are reproducing a known command line.

Advanced submit can pass MMseqs2 subcommands and options directly for custom search/cluster pipelines.
Use canonical configuration for common MSA/search/cluster tasks.

curl -X POST https://subseq.bio/api/v1/job/submit \
  -H "Authorization: Bearer <api_key>" \
  -F program=mmseqs2 \
  -F 'params={"mode":"msa","query_fasta":"query.fasta","sensitivity":"Standard"}'