Figure 3
Experimental validation of the utility of the expertise-driven gene discovery approach. (A) Overlapping TF networks of Gata4 and Tbx20 in cardiac fibroblasts generated by VISIONET, with a filter applied to blur out all gene with Log FC(heart/tail) value between −4 and 4. Node colours were determined according to the heart/tail fibroblast fold-change obtained from microarray data [13]. Gata4, Tbx20, and Aldh1a2 were labelled and enlarged, and other node labels were omitted, for improved visualisation. Squares indicate TFs and circles indicate target genes. The full list of differentially-expressed genes is shown in Additional file 1: S1. Log FC(heart/tail): Log2 of the fold change between heart and tail fibroblast expression. The Venn diagram shows the number of targets (based on ChIP-Seq peaks) of the Tbx20 and Gata4. (B) qPCR validation that the Aldh1a2 gene is uniquely up-regulated in cardiac fibroblasts. Means and standard deviations (n = 3) are shown, and (**) indicates p-value < 0.01 (unpaired t-test). (C) Execution time (seconds) precise to 1 decimal place of the VISIONET web service for the Tbx20-Gata4 network using different layout algorithms. All algorithms were implemented in the same programming language and tested on the same computational hardware for comparison consistency.