“…Given a set of DNA sequences, these programs search for characteristic motifs using two major approaches: profile-based and consensus-based methods. In particular, the recent increase in data size due to the ChIP-Seq technique led to the development of methods that can accept greater than thousands of DNA sequences (Sharov and Ko, 2009;Li, 2008;Heinz et al, 2010;Kulakovskiy et al, 2010;Bailey, 2011;Machanick and Bailey, 2011;Reid and Wernisch, 2011;Ma et al, 2012;Hartmann et al, 2013). Most of these software programs focused on reductions in computational time, for example, by subsampling input data (e.g., MEME-ChIP (Machanick and Bailey, 2011)), accelerating expectation-maximization steps in profile optimization (e.g., ChIPMunk (Kulakovskiy et al, 2010) and STEME (Reid and Wernisch, 2011), which use a greedy approach and suffix array, respectively), or using enriched sequences as starting points for the motif search (e.g., DREME (Bailey, 2011), cERMIT (Georgiev et al, 2010), and HOMER (Heinz et al, 2010)).…”