Extracting Common Motifs under the Levenshtein Measure: Theory and Experimentation

Using our techniques for extracting approximate non-tandem repeats[1] on well constructed maximal models, we derive an algorithm to find common motifs of length P that occur in N sequences with at most D differences under the Edit distance metric. We compare the effectiveness of our algorithm with the more involved algorithm of Sagot[17] for Edit distance on some real sequences. Her method has not been implemented before for Edit distance but only for Hamming distance[12, 20]. Our resulting method turns out to be simpler and more efficient theoretically and also in practice for moderately large P and D.