Fig. 2From: A semi-parametric statistical model for integrating gene expression profiles across different platformsVenn diagram for the detected DEGs reported by different methods in the real data-based simulation. Expression levels of 10000 genes were simulated, among them 30Â % genes are up regulated, 30Â % genes are down regulated and 40Â % genes are non-DE genes. After removing genes with low counts, 6454 genes remain. The Venn diagrams (from left to right) show the intersection between true DEGs and top 10Â %, 20Â % and 30Â % genes ranked by DEseq, eBayes and our method, respectively. At 10Â % cutoff, the detected true positives are 994/3844 = 25.8Â % (our), 980/3844 = 25.4Â % (DEseq) and 988/3844 = 25.7Â % (eBayes), respectively. At 20Â % cutoff, the detected true positive are 1922/3844 = 50.0Â % (our), 1858/3844 = 48.3Â % (DEseq) and 1863/3844 = 48.5Â % (eBayes), respectively. At 30Â % cutoff, the detected true positive are 2626/3844 = 68.3Â % (our), 2508/3844 = 65.4Â % (DEseq) and 2519/3844 = 65.9Â % (eBayes)Back to article page