Definition and identification of homology domains

A method is described for identifying and evaluating regionsof significant similarity between two sequences. The notionof a ‘homology domain’ is employed which definesthe boundaries of a region of sequence homology containing noinsertions or deletions. The relative significance of differentpotential homology domains is evaluated using a non-linear similarityscore related to the probability of finding the observed levelof similarity in the region by chance. The sensitivity of themethod is demonstrated by simulating the evolution of homologydomains and applying the method to their detection. Severalexamples of the use of homology domain identification are given. Received on July 29, 1987; accepted on November 15, 1987