EDAM: An Efficient Clique Discovery Algorithm with Frequency Transformation for Finding Motifs

Finding motifs in DNA sequences plays an important role in deciphering transcriptional regulatory mechanisms and drug target identification. In this paper, we propose an efficient algorithm, EDAM, for finding motifs based on frequency transformation and Minimum Bounding Rectangle (MBR) techniques. It works in three phases, frequency transformation, MBR-clique searching and motif discovery. In frequency transformation, EDAM divides the sample sequences into a set of substrings by sliding windows, then transforms them to frequency vectors which are stored in MBRs. In MBR-clique searching, based on the frequency distance theorems EDAM searches for MBR-cliques used for motif discovery. In motif discovery, EDAM discovers larger cliques by extending smaller cliques with their neighbors. To accelerate the clique discovery, we propose a range query facility to avoid unnecessary computations for clique extension. The experimental results illustrate that EDAM well solves the running time bottleneck of the motif discovery problem in large DNA database.