RAPID AND ACCURATE ESTIMATES OF STATISTICAL SIGNIFICANCE FOR SEQUENCEDATA-BASE SEARCHES

A central question in sequence comparison is the statistical significa nce of an observed Similarity. For local alignment containing gaps to optimize sequence similarity this problem has so far not been solved m athematically. Using as a basis the Chen-Stein theory of Poisson appro ximation, we present a practical method to approximate the probability that a local alignment score is a result of chance alone. For a set o f similarity scores and gap penalties only one simulation of random al ignments needs to be calculated to derive the key information allowing us to estimate the significance of any alignment calculated under thi s setting. We present applications to data base searching and the anal ysis of pairwise and self-comparisons of proteins.