Fig. 4

Distribution of AUC values for classification of tumour and normal samples. The red curve corresponds to RNA-seq data and the blue to HTA data for protein coding mRNA (a) and lncRNA (b). More genes with high AUC, and therefore higher predictive power, were seen for the HTA platform. Fluctuations of RNA-seq distribution for low AUC values are artefacts linked to the limited number of samples