SRGnet (acronym of the Synergistic Response to Gene mutations network) is an R package and available through the Bioconductor repository: https://www.bioconductor.org/packages/SRGnet. 2.6. Data
Polysomal RNA expression was previously measured in young adult mouse colonic epithelium (YAMC) in four conditions: YAMC control, mutant p53-expressing (mp53), activated Ras-expressing, and both mutant genes (mp53/Ras) cells using Affymetrix GeneChip® Mouse Genome 430 2.0 Arrays (McMurray, et al., 2008). The raw data are available via the Gene Expression Omnibus with accession number GSE9199. We processed the transcriptomics data using Robust Multi-array Average (RMA) with custom chip description file containing layout information …show more content…
The enrichment of Role of Deleted in Colorectal Carcinoma (DCC) in regulating apoptosis signal transduction pathway was an important finding that indicated the role of Dapk1 in colorectal carcinoma through regulation of ligand-independent caspase activation (Forcet, et al., 2001). The pathways related to Notch signaling were the next significantly enriched biological pathways with practical implications in targeting cancer cells because of crosstalk with different developmental signaling pathways (Eberl, et al., 2012; Takebe, et al., 2015). Direct p53 effectors pathway was significantly enriched with several CRGs including Pmaip1, Scn3b, Perp, Rb1, Fas and highlighted the roles of CRGs in p53-dependent apoptosis (McMurray, et al., 2008). The tumorigenesis activity of p53 is mediated by Egr1 signaling which enriched as TF of 10.2% of genes in the SRG network (Figure 4C) (Baron, et al., 2005). SP4 transcription factor (TF) was enriched as regulators of almost 11.8 percent of genes in the SRG network (Figure 4C). SP4 is a tumor suppressor gene and is known to be overexpressed in colon cancer (Li, et al., …show more content…
Validation of interactions
We evaluated the robustness of gene-gene interactions in SRG network based on 10-fold cross validation and frequency of their presence in the inferred networks. We hypothesized that if an interaction between pairs of genes is strongly supported by the data, its inference should not be strongly effected by a slightly different training data set. As results, most of interactions in the network of SRGs ranked as highly stable (Figure 5) (Supplementary file 2, Table 3).
In next step, we compared 10 gene regulatory network inference tools based on the same methodology, and SRGnet, Clr, Genie3, mmet, mmetb, pcit, and zscore showed better performance than other tools (Figure 5) (Supplementary Table