Promoter recognition using dinucleotide features : a case study for E.Coli

Promoter recognition is based upon two complementary methods, a motif based method and a global signal based method. The literature is abound with motif search methods. But as the motifs of a promoter are consensus patterns of very short length and the chance of finding putative promoters is high, global feature methods gain importance. In this paper a simple global feature extraction method is proposed for the recognition of sigma-70 promoters in E.coli. It is shown that a simple feed forward neural network classifier achieves a precision of nearly 80% in contrast to the high end classifiers and heavy features proposed in the literature achieving a similar performance. Additionally, a scheme is proposed for locating promoter regions in a given DNA segment.