Local multiple alignment by consensus matrix.

A new algorithm for aligning several sequences based on the calculation of a consensus matrix and the comparison of all the sequences using this consensus matrix is described. This consensus matrix contains the preference scores of each nucleotide/amino acid and gaps in every position of the alignment. Two modifications of the algorithm corresponding to the evolutionary and functional meanings of the alignment were developed. The first one solves the best-fitting problem without any penalty for end gaps and with an internal gap penalty function independent on the gap length. This algorithm should be used when comparing evolutionary-related proteins for identifying the most conservative residues. The other modification of the algorithm finds the most similar segments in the given sequences. It can be used for finding those parts of the sequences that are responsible for the same biological function. In this case the gap penalty function was chosen to be proportional to the gap length. The result of aligning amino acid sequences of neutral proteases and a compilation of 65 allosteric effectors and substrates of PEP carboxylase are presented.