Figure 6

Comparison of different approaches. GO analysis of path enrichment in dynamic networks constructed by DREM 2.0 for the asbestos data set. The enrichment of GO terms for all paths, after Bonferroni multiple testing correction, is depicted. Three different learning scenarios are compared: construction without any TF input just using the mRNA expression data (mRNA exp), construction using protein-DNA predicted binding events (mRNA exp+TF), and construction using protein-DNA predicted binding events and the new TF-expression scaling method (mRNA exp+TF exp). A) Comparison of the enriched GO terms with corrected p-value below 0.05 for each method shown as a Venn diagram. B) Display of GO terms that are uniquely identified by each method. Leveraging the TF-expression level improves the GRN construction in addition to using the protein-DNA interaction data.