Eukaryotic promoter prediction based on relative entropy and positional information

Abstract

The eukaryotic promoter prediction is one of the most important problems in DNA sequence analysis, but also a very difficult one. Although a number of algorithms have been proposed, their performances are still limited by low sensitivities and high false positives. We present a method for improving the performance of promoter regions prediction. We focus on the selection of most effective features for different functional regions in DNA sequences. Our feature selection algorithm is based on relative entropy or Kullback-Leibler divergence, and a system combined with position-specific information for promoter regions prediction is developed. The results of testing on large genomic sequences and comparisons with the PromoterInspector and Dragon Promoter Finder show that our algorithm is efficient with higher sensitivity and specificity in predicting promoter regions.