Automat is devoted to scan databases of nucleic acid or amino acid sequences in order to retrieve sequences distantly related to the query.

Like BLAST, it is based upon an automaton that splits the query sequence into words of increasing length.

For proteins, amino acids can be gathered into classes at will. The default length of the word that trigger a match is 3 amino acids for proteins and 5 nucleic acids for DNA. The total length of the local alignment is fixed at 8 by default for proteins and 25 for DNA, and gaps are allowed on this fragment for a default value of 20 for both proteins and DNA.

Automat is devoted to search for local anchor regions. An algorithm of suppression of redundancies in the output listing is available in order to make the results easier to interpret.