Fig. 4

Performance characteristics of the “Classifier” module for host transcript expression profiling. a Published RNA-seq data from a commercially available RNA standard (MAQC, Additional file 1: Table S12) were analyzed by Taxonomer, Sailfish, and Cufflinks and estimated transcript expression was compared to data obtained by quantitative PCR (qPCR). Gene-level Pearson and Spearman correlation coefficients for RNA-seq vs. qPCR were 0.85 and 0.84 for Taxonomer, 0.87 and 0.86 for Sailfish, and 0.80 and 0.80 for Cufflinks, respectively. b Application of Taxonomer to metagenomic RNA-seq data from routine respiratory samples from patients with influenza infection (n = 4). c Classification of viral sequencing reads by Protonomer and typing of this strain as influenza A(H1N1)pdm09 (top right sample from a). d Differential gene-level mRNA expression profiles from four patients with influenza A virus compared to asymptomatic controls (n = 40; top 50 differentially expressed genes are shown). Expression profiles for 17 genes were significantly higher in influenza-positive patients (Additional file 1: Table S5). e Expression profiles for the 17 most differentially expressed genes differentiate cases from controls (principal component analysis, PC1 and PC2 explaining 93.8 % of the total variance). f Normalized expression levels for individual patients of seven of the top 17 genes. Gene ontology assignments for enrichment of biological processes (g) and molecular functions (h) are shown