SCORE: A computational approach to the identification of cis-regulatory modules and target genes in whole-genome sequence data

Abstract

A large fraction of the information content of metazoan genomes resides in the transcriptional and posttranscriptional cis-regulatory elements that collectively provide the blueprint for using the proteincoding capacity of the DNA, thus guiding the development and physiology of the entire organism. As successive whole-genome sequencing projects--including those of mice and humans-are completed, we have full access to the regulatory genome of yet another species. But our ability to decipher the cis-regulatory code, and hence to link genes into regulatory networks on a global scale, is currently very limited. Here we describe SCORE (Site Clustering Over Random Expectation), a computational method for identifying transcriptional cis-regulatory modules based on the fact that they often contain, in statistically improbable concentrations, multiple binding sites for the same transcription factor. We have carried out a Drosophila genomewide inventory of predicted binding sites for the Notch-regulated transcription factor Suppressor of Hairless [Su(H)] and found that the fly genome contains highly nonrandom clusterings of Su(H) sites over a broad range of sequence intervals. We found that the most statistically significant clusters are very heavily enriched in both known and logical targets of Su(H) binding and regulation. The utility of the SCORE approach was validated by in vivo experiments showing that proper expression of the novel gene Him in adult muscle precursor cells depends both on Su(H) gene activity and sequences that include a previously unstudied cluster of four Su(H) sites, indicating that Him is a likely direct target of Su(H).R ealizing the full promise of whole-genome sequencing projects depends on our ability to read and understand the tremendous informational richness contained therein. Computational methods for predicting novel protein coding genes in whole-genome sequence data are quite advanced, and various strategies for recognizing transcription units that generate noncoding RNAs are also available. However, these techniques address only part of the informational content of the genome. The complex blueprint that controls the utilization of the coding information in DNA is contained in the huge number of cis-regulatory elements, both transcriptional and posttranscriptional, that surround and invade the transcribed part of the genome. But it is clear that we are in our infancy in learning how to read the regulatory genome and thus decipher the blueprint.Here we describe SCORE (Site Clustering Over Random Expectation), a computational method for identifying potential cisregulatory modules and the target genes they serve. Transcriptional enhancer elements are generally quite compact, and they frequently include closely spaced binding sites for the same or multiple transcription factors (1). SCORE is designed to detect and statistically evaluate these structural features in whole-genome sequence data, and thus to reveal previously unrecognized enhancers. A conceptually similar method has been describ...

Neural precursor-specific expression of multiple Drosophila genes is driven by dual enhancer modules with overlapping function

“…Statistically significant clustering of TF binding sites has proven to be a valuable criterion for the identification of functional CRMs (12,25). We applied a two-motif version of the SCORE computational technique (12) to recognize several NP-specific enhancers in the fly genome. In the case of three genes (phyl, sens, and nvy), the newly identified NP module is distinct from another one previously reported, establishing the existence of a dual-enhancer configuration.…”

“…S1) suggested that additional SOP-specific [or, more generally, neural precursor (NP)-specific] enhancer modules might be identifiable on this basis. Accordingly, we used the SCORE computational method (12) to survey the Drosophila genome for statistically significant coclustering of P and R motifs. We detected such clusters in association with several known NP-expressed genes, including phyllopod (phyl), senseless (sens), and nervy (nvy) (Fig.…”

“…Here we have used this scenario as the basis of a computational approach to identifying, in the Drosophila genome, NPspecific enhancer modules associated with NP-expressed genes (12). We searched for statistically significant motif clusters that include predicted binding sites for both proneural proteins (P) and Hes repressors (R), and successfully detected "P+R" enhancers in the vicinity of multiple genes with NP-specific expression.…”

Cooperative assembly of higher-order Notch complexes functions as a switch to induce transcription

“…One clue is found in the promoter regions of a number of well characterized Notch-responsive genes (19), such as the hairy/enhancerof-split genes in Drosophila and a number of their mammalian homologues, which contain dual ''sequence-paired'' binding sites (SPSs) (20,21). SPSs consist of two CSL-binding sites oriented head-to-head that typically are separated by 16 or 17 nt, depending on the species examined [supporting information (SI) Fig.…”

Identifying regulatory elements in eukaryotic genomes

“…Homotypic clusters have been widely studied in Drosophila [157][158][159], but are yet relatively unexplored in mammalian genomes. A large fraction of methods use a set of elements believed to be functional in a particular process or cell type and train a model based on the frequencies and relative distributions of motifs within them.…”