Genome-wide prokaryotic promoter recognition based on sequence alignment kernel

Abstract

In this paper an application of Sequence Alignment Kernel for recognition of prokaryotic promoters with transcription start sites (TSS) is presented. An algorithm for computing this kernel in square time is described. Using this algorithm, a "promoter map" of E.coli genome has been computed. This is a curve reflecting the likelihood of every base of a given genomic sequence to be a TSS. A viewer showing the likelihood curve with positions of known and putative genes and known TSS has also been developed and made available online. Although the visual analysis of the promoter regions is very intuitive, we propose an automatic genome-wide promoter prediction scheme that simplifies the routine of checking the whole promoter map visually. Computational efficiency and speed issue are also discussed. © Springer-Verlag Berlin Heidelberg 2003.