In this paper we propose a novel approach for identification of generic motifs in an integrated manner by introducing the notion of submotifs. We formulate the motif finding problem as a constrained submotif pattern mining and present an algorithm called SPACE for identifying motifs that may contain spacers. When spacers are present, we show that the algorithm can identify motifs where 1) the spacers may be of varying lengths, 2) the number of motif segments may be unknown, and 3) the lengths

more »

... motif segments may be unknown. We perform rigorous experiments with the Motif Assessment Benchmarks by Tompa et al., and observe that our algorithm overall is able to outperform all popular algorithms tested so far, with significant improvements on sensitivity and specificity.