Detection and identification of cis-regulatory elements using change-point and classification algorithms

Abstract

Background Transcriptional regulation is primarily mediated by the binding of factors to non-coding regions in DNA. Identification of these binding regions enhances understanding of tissue formation and potentially facilitates the development of gene therapies. However, successful identification of binding regions is made difficult by the lack of a universal biological code for their characterisation. Results We extend an alignment-based method, , and identify clusters of biological significance, through ontology and de novo motif analysis. Further, we apply a Bayesian method to estimate and combine binary classifiers on the clusters we identify to produce a better performing composite. Conclusions The analysis we describe provides a computational method for identification of conserved binding sites in the human genome and facilitates an alternative interrogation of combinations of existing data sets with alignment data.