Fig. 3
Differentially expressed genes identified by the platforms. Evolution of the number of significant genes identified with variable FDR thresholds (a), using edgeR and limma with voom correction for analysis of RNA-seq data, and using limma for HTA data. Solid lines show unpaired analyses, while dotted lines show analyses paired by patient. Differentially expressed protein coding mRNAs (b) and lncRNAs (c) were obtained by unpaired differential expression analysis using edgeR for RNA-seq and limma for HTA (FDR < 0.01) and represented as proportional Euler-Venn diagrams. The lists of differentially expressed genes were confirmed by the top 25% significant genes detected in the LUSC-TCGA dataset: 4569 protein coding genes (FDR < 10−18) and 111 lncRNAs (FDR < 10−8) were used. Evolution of Jaccard index for coding mRNAs with variable FDR thresholds (d) between the two platforms (violet) shows a monotonic behaviour. Similarity between the TCGA validation gene list and each of the platforms– RNA-seq (red) and HTA (blue) showed a slight outperformance of HTA (marked by an arrow)