[TOC]
Abstract
There is a critical need for standard approaches to assess, report and compare the technical performance of genome-scale differential gene expression experiments
Here we assess technical performance with a proposed standard ‘dashboard’ of metrics derived from analysis of external spike-in RNA control ratio mixtures
These control ratio mixtures with defined abundance ratios enable assessment of diagnostic performance of differentially expressed transcript lists, limit of detection of ratio (LODR) estimates and expression ratio variability and measurement bias
Ratio measurement variability and bias are also comparable among laboratories for the same measurement process.We observe different biases for measurement processes using different mRNA-enrichment protocols
Intro
Ratios of mRNA transcript abundance between sample types are measures of biological activity. These measurements of differential gene expression are important to underpin new biological hypotheses and to support critical applications such as selection of disease classifiers and regulatory oversight of drug therapies.
Controls and associated ratio performance metrics are essential to understand the reproducibility and validity of differential expression experimental results. External spike-in control ratio measurements can serve as a truth set to benchmark the accuracy of endogenous transcript ratio measurement
Method validation of differential expression experiments based on these ERCC controls is the focus of this work. This validation supports comparisons across experiments, laboratories, technology platforms and data analysis methods3–7. In any differential expression experiment, with any technology platform, a pair of ERCC control ratio mixtures can be added (‘spiked’) into total RNA samples such that for each ERCC control the relative abundance of the control between samples (ratio) is either of known difference (a true-positive control) or the same (a true-negative control).
These ratio performance measures include diagnostic performance of differential expression detection with receiver operating characteristic (ROC) curves and area under the curve (AUC) statistics, limit of detection of ratio (LODR) estimates and expression ratio technical variability and bias
