Reflection back to the OI MET gene expression signature Keeping in mind that the OI MET TF network is neces sarily simplistic, this network is strongly consistent with the hypothesis that the OVOLs regulate MET in concert with the other four TFs. However, since the roles of the other four tech support TFs were suggested by enrichment of annota tion in the OI MET signature gene set, we hypothesized that the effects of these TFs from the OI MET TF model are consistent across the larger OI MET signature set. We also observed that, in addition to the four TFs iden tified by ConceptGen, MAFF, ATF3, MYC, MYB, and IRF9 could be important in this regulatory cascade. Using GGAs MatInspector function, we searched the 4,102 promoters from the OI MET gene set, looking for individual binding sites for these promoters.

Based on the number of sequences with one or more binding sites for each of these TFs, and comparing to the frequency expected for all promoters, we find that NFKB, MYC, and to a lesser extent, MAFF motifs are over represented in these promoters. Note that, while these motifs are over represented, the modest values of over representation make their biological relevance subject to interpretation. Equally, as noted below, the presence of single motifs is not a strong indicator of regulatory control. The proportion of promoters with the other motifs is not significantly different from the proportion expected for a random set of promoters at a significance threshold of p value 0. 05. Since TFs generally work in pairs or modules, we searched the 4,102 promoter sequences for all pairs of motifs derived from these individual motifs using GGAs RegionMiner module.

RegionMiner compared the proportion of promoters with each motif pair in the 4,102 promoters versus the proportion of promoters with the motif pair in all GGA promoters. This is the observed enrichment in Table 6. For comparison, we calculated the expected representation in this group of 4,102 promoters as Drug_discovery the product of fold enrichment for the first motif x the fold enrichment for the second motif. This is the value expected if the motifs were randomly distributed across the 4,102 promoters. For almost all of these TF pairs, we found approximately the expected number of promoters with the motif pair. However, the VAP1F/VEBOX motif pair, corresponding to the AP1/MYC TF pair, showed 1.

38 fold enrichment relative to all promoters in the RegionMiner search. Based on our calculation, we would have expected only 1. 07 fold enrichment. Leukemia This difference is the largest in our dataset and is significant at the ��2 p value 0. 01 level. Finding a much greater proportion of promoters with the motif pair than expected by chance suggests that coopera tive regulation by AP1 and MYC could be important in the downstream cascade of gene expression regulating MET.