Comprehensive multi-center assessment of accuracy, reproducibility and bias of small RNA-seq methods for quantitative miRNA profiling

RNA-seq is increasingly employed for quantitative profiling of small RNAs (e.g., microRNAs, piRNAs, snoRNAs) in diverse sample types including isolated cells, tissues and cell-free biofluids. The accuracy and reproducibility of the multiple small RNA-seq library preparation methods in use, however, have not been systematically assessed. We report systematic results obtained by a consortium of nine labs that independently sequenced reference, ‘ground truth’, samples of synthetic small RNAs and human plasma-derived RNA. Three commercially available library preparation methods employing adapters of defined sequence and six methods using adapters with degenerate bases were assessed. Both protocol- and sequence-specific biases were identified, including biases that reduce the ability of small RNA-seq to accurately measure adenosine-to-inosine editing in microRNAs. We report that these biases were mitigated by library preparation methods that incorporate adapters with degenerate bases. MicroRNA relative quantification between samples using small RNA-seq was found to be accurate and reproducible across laboratories and methods.