Application of Random Forest Classifier to DNA Promoter Site Data

In this paper we address the problem of promoter site recognition in eukaryotes DNA sequence. We apply a novel approach of the recently introduced random forest classifier that has been shown to out perform neural networks. Preliminary results are presented of our long term effort of developing efficient promoter site recognition software module. Our approach involves combination of several advanced mathematical techniques to develop an efficient module in be incorporated into gene recognition software. Gene recognition is essential to understanding existing and future DNA sequence data. During transcriptional initiation, there is a large variety of transcription factors interacting and cooperating in promoter regions in complex ways. To answer the question of how genetic information is processed, promoter identification becomes a necessary step, especially in eukaryotes in which the promoters are involved in various biochemical processes. Developing computational methods to find promoter sequence patterns is therefore vital for achieving the goals of for example the Human Genome Project.