HMMER: biosequence analysis using profile hidden Markov models

HMMER is used for searching sequence databases for sequence homologs, and for making sequence alignments. Its methods use probabilistic models called profile hidden Markov models (profile HMMs).

You can use hmmbuild and hmmsearch to build and search with your own profiles of conserved protein domains or DNA elements starting from your own multiple sequence alignments. HMMER is also often used together with an existing curated profile database, such as Pfam or many of the databases that participate in Interpro: you can scan a sequence against a profile database with hmmscan. HMMER can also start with your query sequences, not just profiles, just like BLAST. You can search a protein query sequence against a database with phmmer, or do an iterative search with jackhmmer.

HMMER is designed to detect remote homologs as sensitively as possible, relying on the strength of its underlying probability models. When profile HMMs were first introduced, this strength came at significant computational expense, but as of HMMER3, HMMER is essentially as fast as BLAST.

HMMER can be downloaded and installed as a command line tool, and a HMMER search server is hosted at the European Bioinformatics Institute.