Practical strategies for discovering regulatory DNA sequence motifs

Abbreviations: A, adenine; C, cytosine; ChIP, chromatin immunoprecipitation; EM, expectation maximization; G, guanine; ROC, receiver operating characteristic; ROC-AUC, area under the receiver operating characteristic curve; T, thymine Many functionally important regions of the genome can be recognized by searching for sequence patterns, or "motifs." Moving forward, we believe that cross-validated hypothesis testing and regression-based approaches will prove to be particularly valuable, as they combine the data-mining capabilities of classic motif discovery programs with a framework that offers an intuitive interpretation of the motifs.