Parallel Discovery of Transcription Factor Binding Sites

Transcription factor binding sites (TFBSs) are cis-acting control elements that regulate gene expression upon binding to their corresponding transcription factors (TFs). Genome wide discovery of transcription factor binding sites (TFBSs) would aid in understanding gene expression. Various experimental techniques are currently used for TFBS discovery, but are limited in scale and speed. Computational approaches have proven to be fast and reliable alternatives for discovering TFBS on a whole genome scale and offer several advantages to experimental techniques. However, algorithms for TFBS discovery are highly compute-intensive. This has so far limited their application to relatively small problem sizes. This paper presents the design of a parallel TFBS discovery algorithm. Its implementation on a high performance compute cluster leads to significant runtime savings