THE STATISTICAL SIGNIFICANCE OF NUCLEOTIDE POSITION-WEIGHT MATRIX MATCHES

Motivation: To improve the detection of nucleotide sequence signals (e .g. promoter elements) by position-weight matrices (PWM) using the con cept of statistically significant matches. Results: The Mksite program was originally developed for analyzing protein sequences. We report N Mksite, a new version adapted to the processing of nucleotide sequence s. NMksite creates PWM from nucleotide sequence block alignments or oc currence tables using three weight computation schemes. An original fe ature of NMksite is the numerical computation of the statistical signi ficance of PWM matches. The utility of this concept is demonstrated in the context of the prediction of splice sites and promoter regions. A vailability: Mksite and other components of the MODEST (Motif DEsign a nd Search Tool) package (written in C/Unix) are available at http://ig s-sewer.cnrs-mrs.fr.