Promoter Analysis with Wavelets and Support Vector Machines

AbstractPromoters are key control regions for the transcription regulation of genes, usually lying upstream of the genes they control. Promoter prediction is worthwhile not only for the detection of orphan genes but also for understanding the mechanisms that regulate gene expression. Promoter prediction therefore remains one of the primary challenges subjects in bioinformatics in the post-genome era. Many methods are used for promoter prediction, such as the presence of the CpG islands, sequence motifs of transcription factor binding sites, and the statistical and chemo- physical properties in the vicinity of transcription start sites. Among these strategies, we have focused on a method which employs wavelet analysis and support vector machine for promoter prediction. The wavelet analysis is based on localized wave packets characterized by both a range of frequency and a location. In our scheme, information from promoter and non-promoter regions is converted to wavelet space as a positive and a negative set, respectively, and the 2 sets are subsequently used to train a support vector machine. Finally, the support vector machine is utilized for promoter prediction. In this study, we improved the coding method of our prediction strategy and analysed a new set of test data