Research Paper Volume 15, Issue 2 pp 472—491

Figure 2. Detailed analysis of the Interaction group. (A) Left: Euler diagram of significant genes (also in Figure 1), focusing on the 1576 DEGs rated significant by the two-way ANOVA ‘interaction’ term. Center: A grid of all possible temporal patterns of expression for Vehicle samples (horizontal) and Myriocin samples (vertical) was used to count the number of Interaction DEGs assigned to each pattern. Results are shown as a heatmap with green highlighting. There are five patterns of interest (labels indicated) that each showed >100 DEGs (total results for this grid are available in Supplementary Table 1, tab called ‘I Intersections’). Right: The number of interaction-significant DEGs within each of these five patterns of interest are graphed according to whether Myr increased (orange) or decreased (blue) expression levels compared to vehicle-treated samples. Myr’s effect was predominantly to increase expression. (B) Left: Templates of expression for the five Myr-increased patterns of interest. Center: Heatmaps showing signal intensities for representative genes within each statistically identified pattern. Right: average Z-scored expression levels for all genes assigned to each selected pattern (functional overrepresentation analysis results for these patterns are presented in Table 2). (C) Left: The analysis of interaction-significant DEGs, particularly in 2B, suggested a simplified and more general trend among the genes found to be significant by the two-way ANOVA’s Interaction term. That is, Myr treatment appeared to increase expression at time points 5 and 6 regardless of the temporal pattern of expression. To test this, we plotted the log 2-fold changes of Myr treatment/vehicle treatment at each time point, and color-coded those results if they were significantly (p ≤ 0.05, pairwise Fisher’s LSD) downregulated (blue) or upregulated (orange). As expected, the predominant effect was upregulation at time points 5 and 6. Right: Functional overrepresentation analysis. Columns- Tot: total number of genes in the dataset assigned to GO term; Sig: number of total genes that were significant and were assigned to the indicated pattern of expression; Exp: number of genes expected to be found in that pathway for that pattern of expression by chance; Ratio: Sig/Exp; p-value: overrepresentation analysis ([46]); FDR: multiple testing adjusted ORA p-alue according the BH procedure ([47]; Note- complete list of all pathways, and the genes assigned to them, is provided in Supplementary Table 2, Webgestalt patterns combined tab).