A METHOD FOR IDENTIFYING SPLICE SITES AND TRANSLATIONAL START SITES IN EUKARYOTIC MESSENGER-RNA

This paper describes a new method for determining the consensus sequen ces that signal the start of translation and the boundaries between ex ons and introns (donor and acceptor sites) in eukaryotic mRNA. The met hod takes into account the dependencies between adjacent bases, in con trast to the usual technique of considering each position independentl y. When coupled with a dynamic program to compute the most likely sequ ence, new consensus sequences emerge. The consensus sequence informati on is summarized in conditional probability matrices which, when used to locate signals in uncharacterized genomic DNA, have greater sensiti vity and specificity than conventional matrices. Species-specific vers ions of these matrices are especially effective at distinguishing true and false sites.