Alignment method for comparing a protein sequence to a protein family, represented by a multiple alignment. It can also be used for sensitive protein database searches. The algorithm is a generalization of the Smith-Waterman algorithm.