Abstract
We present here an efficient fast algorithm that aims at detecting statistically significant motifs in a set of unaligned DNA sequences. This algorithm has a resemblance to the expectation maximization (EM) algorithm, but is derived using heuristics and not probabilistic models. The algorithm is implemented in the Dragon Motif Builder system and evaluated on sets of 11,502 mouse and 18,326 human promoter sequences each of length of 1500 bp. The web-server with this algorithm implemented is freely available for academic and non-profit users via http://research.i2r.a-star.edu.sg/DRAGON/MotiLSearch/.