Truth and Service
Future StudentsStudentsParents and FamilyAlumniFaculty and StaffVisitors
Directory Image
Customer Service
Gene Set Enrichment: Findings

The dynamic Bayesian network tool Banjo was used to construct the gene regulatory networks (see for instance ). To combine the networks from different experimental data set, we first represent each network using an adjacency matrix, and element-wise addition is applied to the adjacency matrices to combine the results from multiple experiments. The final adjacency matrix is then converted back to the format that is acceptable to graph drawing utility GraphViz. For Gene Set Enrichment Analysis, we ranked our genes with their meta P-values and test their enrichment in 4 functionally known gene clusters (e.g. CERS-up, Cdc15, Kap123 and Wos2).

We performed differential meta-analysis for 4940 genes. We reported 43 genes that are stress responsive. For the 43 genes, we reconstructed a gene regulatory network based on the integrative approach described above (Figure 1). Among the genes in the network, some interesting hub nodes such as ssa2 were observed.

4_1

Figure 1. The regulatory network based on the significant genes from differential meta-analysis. The rectangular nodes represent known stress response genes; the double-rimmed nodes are genes from the core stress gene list while single-rimmed ones belong to the Non-core Stress Gene list. The ellipses represent genes that are in neither list. A blue edge represents regulation that is significant in more Elutriation experiments than Cdc25, and a red represents the converse situation. A prominent hub node, ssa2, which is listed neither as core stress gene list nor as non-core stress gene list, is marked with a magenta circle.


Gene Set Enrichment Analysis results are shown in Figure 2.

4_2

Figure 2. Gene Set Enrichment Analysis for known gene sets in S. pombe. The enrichment scores (in green) are computed for four known gene clusters (named within each plot) over the global list of 4940 genes ranked by their differential meta-analysis p-values. The list is depicted from the lowest p-value (i.e. for gene with high expression and low periodicity) at the left end to the highest p- value (the converse scenario) at the right end, and a “hit” for a gene from the chosen cluster is recorded with a black vertical mark.