Pfam is a database of protein families that includes their annotations and multiple sequence alignments generated using hidden Markov models.