What can we do with multiple alignments?
Create (databases of) profiles derived from multiple alignments for protein families
- profile = multiple alignment + observed character frequencies at each position
Search with a sequence against a database of profiles (e.g. PROSITE database)
- faster than sequence against sequence
- gives a more general result (“the input sequence matches globin profile”)
Search with a profile against a database of sequences
- PSI-BLAST : can identify more distant relationships than by normal BLAST search