PMC

Normalized cumulative histograms of mean fold changes in the microarray data set for all genes (open bars), genes where Cluster-Buster detected a cluster of binding sites (gray bars) and after additional filtering with Gene Ontology (closed bars).

Data flow in our method: Using the Cluster-Buster algorithm, we search for clusters of binding sites in putative promoter regions. The list of genes having a cluster in their promoter are then passed to GOSSIP, which detects association with biological processes using the Gene Ontology. The significantly associated processes are reported.

Black and gray boxes indicate the significantly associated biological processes [FDR ≤ 0.01 and FDR ≤ 0.05, respectively] with (a) the set of transcription factors CREB, CEBP, p50/p65, Sp-1, ETS and AP1 as predicted by our method; (b) up-regulated genes upon LPS stimulation in monocytes (microarray data from the alliance for cellular signaling).

Results for the set of transcription factors Mef-2, Myf and TEF, known to regulate the expression of muscle-specific genes (). The black and gray boxes correspond to significantly overrepresented biological processes of the Gene Ontology within the predicted target genes [thresholds of FDR ≤ 0.01 and FDR ≤ 0.05, respectively]. The diamond shows the root node for biological processes. (a) An illustration of the complexity of the analysis: overrepresented terms drawn in the context of all 655 terms assigned to the genes with predicted clusters of binding sites. (b) A fragment emphasizing all significantly overrepresented terms.