[Computation of the probabilities of families of biological sequences]

An algorithm for computing the probabilities of families of biological sequences is presented. The algorithm is applicable to many problems of bioinformatics, in particular, for computing the sensitivity of seeds while searching for local similarities in genomes and while estimating the reliability of search for clusters of regulatory sites. This algorithm can be also used for distribution of probabilities described by different models, for example, Bernoulli, Markov, and hidden Markov models. The algorithm is based on the description of probabilities distribution as well as on the description of the family of sequences using the finite automaton. Then the problem of calculation of probabilities is reduced to computation of the proper generalized statistical sum. The algorithm can be applied not only to biological sequences but also to symbol sequences of any origin.