Identification of Over-Represented Combinations of Transcription Factor Binding Sites in Sets of Co-Expressed Genes

Abstract

Transcription regulation is mediated by combinatorial interactions between diverse trans-acting proteins and arrays of cis-regulatory sequences. Revealing this complex interplay between transcription factors and binding sites remains a fundamental problem for understanding the flow of genetic information. The oPOSSUM analysis system facilitates the interpretation of gene expression data through the analysis of transcription factor binding sites shared by sets of co-expressed genes. The system is based on cross-species sequence comparisons for phylogenetic footprinting and motif models for binding site prediction. We introduce a new set of analysis algorithms for the study of the combinatorial properties of transcription factor binding sites shared by sets of co-expressed genes. The new methods circumvent computational challenges through an applied focus on families of transcription factors with similar binding properties. The algorithm accurately identifies combinations of binding sites over-represented in reference collections and clarifies the results obtained by existing methods for the study of isolated binding sites.