GENOME STRUCTURE AND FUNCTION IDENTIFICATION OF CODING REGIONS IN GENOMES OF

Motivation: It is known that the protein-coding regions exhibit a lower degree of compositional variation than the intergenic regions or introns. In this study we tried to consider an inverse problem, that is to segment the sequence of a complete genome into regions of homogeneous composition and then to compare those regions with various functional units of a genome. Results: We applied our program of compositional segmentation into compositionally homogeneous domains to several sequences of complete chromosomes of lower eukaryotes. We have identified DNA segments of the three main types: a) coding segments, b) long repeats, c) AT-rich intergenic segments. After filtering out the AT-rich regions and the repeats, we have obtained the quality of the prediction of coding sequences at the nucleotide level comparable to that of specialized gene finding programs. Several additional candidate coding regions have been identified