Identifying prokaryotic promoter sequences is notoriously difficult and for most sequenced bacterial genomes the promoter sequences are still unknown. Since experimental analysis trails behind sequencing, genome-wide computational promoter discovery is often the only realistic way to discover these sequences in newly sequenced bacterial genomes. However, genome-wide samples for promoter discovery may be very large and corrupted complicating promoter discovery. We discuss three aspects of genome-wide promoter discovery: sample generation, signal finding algorithms, and scoring signals. We applied our new MITRA algorithm to analyze samples of divergent and convergent genes in 20 bacterial genomes and found strong putative dyad signals in 17 out of the 20 genomes. Moreover, in 12 out of 20 genomes the found signals are identical or similar to the known regulatory patterns (Pribnow-Gilbert boxes and CRP binding sites). Since many of putative signals correspond to previously known elements of bacterial transcriptional regulation, the remaining discovered signals are good candidates for unknown regulatory elements.
|Number of pages||12|
|Journal||Pacific Symposium on Biocomputing. Pacific Symposium on Biocomputing|
|Publication status||Published - 2003|