Identification of cis-acting regulatory elements using orthology biased Gibbs sampling

Many computational pattern searching tools for the discovery of novel, common regulatory elements between co-expressed genes have been developed over the last ten years. However, few approaches attempt to incorporate valuable additional information, such as inter-species conservation, into the prediction process. Orthology biased Gibbs sampler (OrBS) is an expansion on the Gibbs sampler motif discovery approach. I introduce dynamic motif width prediction, a novel convergence detection approach, and a scoring function that incorporates cross-species sequence conservation. The algorithm was refined using the Caenorhabditis elegans X box element and is shown to successfully identify the element in sequence sets with only 33% of X box regulated genes. Using the reported X box consensus, I successfully identified additional genes, like the C. elegans orthologue to human BBS4. OrBS was less successful in the identification of the other C. elegans regulatory elements, such as the PHA-4 binding site and the UNC-86 binding site.