HMMER is an implementation of profile HMM methods for sensitive database 
searches using multiple sequence alignments as queries.

Basically, you give HMMER a multiple sequence alignment as input; it builds 
a statistical model called a "hidden Markov model" which you can then use as 
a query into a sequence database to find (and/or align) additional 
homologues of the sequence family.

Hmmer is developed by Sean Eddy at Washington University School of Medicine.
More information is available at http://hmmer.wustl.edu/