A Statistical Method for Finding Transcription Factor Binding Sites

Understanding the mechanismsthat determine the regulation of geneexpression is an important and challenging problem. A fundamental subproblem is to identify DNA-bindingsites for unknownregulatory factors, givena collection of genesbelieved to be coregulated, and given the noncoding DNA sequences near those genes. Wepresent an enumerative statistical methodfor identifying good candidates for such transcription factor binding sites. Unlikelocal search techniques such as Expectation Maximizationand Gibbs samplers that maynot reach a global optimum,the methodproposed here is guaranteed to produce the motifs with greatest z-scores. Wediscuss the results of experiments in which this algorithm wasused to locate candidatebindingsites in several well studied pathwaysof S. cerevisiae, as well as geneclusters from someof the hybridization microarrayexperiments.

• Type: Conference
• Year: 2000
• Where: ISMB
• Authors: Saurabh Sinha, Martin Tompa
• Keywords: Computational Biology, Enumerative Statistical Methodfor, Expectation Maximizationand Gibbs, ISMB 2000, Noncoding Dna Sequences