Generalized Composite Motif Discovery

This paper discusses a general algorithm for the discovery of motif combinations. From a large number of input motifs, discovered by any single motif discovery tool, our algorithm discovers sets of motifs that occur together in sequences from a positive data set. Generality is achieved by working on occurrence sets of the motifs. The output of the algorithm is a Pareto front of composite motifs with respect to both support and significance. We have used our method to discover composite motifs for the AlkB family of homologues. Some of the returned motifs confirm previously known conserved patterns, while other sets of strongly conserved patterns may characterize subfamilies of AlkB.