Skip to main content
Figure 2 | BMC Research Notes

Figure 2

From: Validation of MIMGO: a method to identify differentially expressed GO terms in a microarray dataset

Figure 2

Validation-flow chart of GSEA + MIMGO. Eighteen row- and 18 column-indices labeled by time-points (0, 7, 14, etc.) in a matrix of step 1 show the time-course microarray data. GSEA determines whether a gene set assigned a GO term is differentially expressed between row- and column-indices in the matrix of step 1. Log2_Ratio_of_Classes is used as a metric for GSEA. In step 2, MIMGO identifies differentially expressed GO terms using the matrix of step 1. To investigate whether the “microarray data with upregulation” identified by GSEA + MIMGO are correct, the results obtained from GSEA + MIMGO are compared with the average expression level of genes annotated with a GO term for each microarray result. In the square below, the average expression of a gene set annotated with a GO term exceeds the sum of its average and standard deviation only at 14 min, 21 min, 70 min, and 77 min. Therefore, these time-points are marked with 1 in “> (avg + std)”. GSEA (Pearson) shows GSEA using Pearson correlation as metrics. Ideal expression pattern in the time-course microarray data (0 min, 7 min, 14 min, etc.) is prepared as a phenotype label. Pearson correlation coefficients are calculated between the ideal expression pattern and actual gene expressions. The GO term associated with a gene set showing a high Pearson correlation is identified as a differentially expressed GO terms. Finally, “GSEA + MIMGO” is compared with GSEA (Pearson) for identification of true differentially expressed GO terms.

Back to article page