Systematic assessment of next generation sequencing for quantitative small RNA profiling: a multiple protocol study across multiple laboratories

Abstract

Small RNA-seq is increasingly being used for profiling of small RNAs. Quantitative characteristics of long RNA-seq have been extensively described, but small RNA-seq involves fundamentally different methods for library preparation, with distinct protocols and technical variations that have not been fully and systematically studied. We report here the results of a study using common references (synthetic RNA pools of defined composition, as well as plasma-derived RNA) to evaluate the accuracy, reproducibility and bias of small RNA-seq library preparation for five distinct protocols and across nine different laboratories. We observed protocol-specific and sequence-specific bias, which was ameliorated using adapters for ligation with randomized end-nucleotides, and computational correction factors. Despite this technical bias, relative quantification using small RNA-seq was remarkably accurate and reproducible, even across multiple laboratories using different methods. These results provide strong evidence for the feasibility of reproducible cross-laboratory small RNA-seq studies, even those involving analysis of data generated using different protocols. (Introduction without separate heading below)RNA-seq using next generation sequencing has been a transformative technology that has been widely used as a method for characterizing the transcriptome in a wide range of biological contexts 1,2 . Applications of RNA-seq fall into two categories: long RNA-seq and small RNA-seq, distinguished not only by the size of the targeted RNAs, but also by the technical methods used and the resulting biases of the different approaches in the quantitative data produced 3 . For example, the production of the libraries for long RNA-seq, by virtue of having sufficiently long target RNA lengths, commonly utilizes primers (e.g., random primers or oligo-dT) for direct generation of cDNA from RNA. However, small RNA-seq library construction methods typically require an RNA ligation or polyA tailing step to overcome the challenge of performing reverse transcription and subsequent PCR-based amplification from extremely short (e.g., 16-30 nt) target RNA sequences.Multiple approaches have been developed to overcome the challenge of uniformly and robustly generating cDNA from small RNAs for the purpose of small RNA-seq library preparation 4-9 . Protocols in use for small RNA-seq therefore vary more widely than those used for long RNA-seq, creating significantly more potential for variation across small RNA-seq results using different library preparation protocols and by different labs. In addition, small RNA-seq is increasingly used to study small RNAs present in very low input concentration samples (e.g., in exosomes and other types of extracellular vesicles (EV) 10-19 , or in RNA-protein complexes present in biofluids 20-26 ). Normalization methods 27-29 developed to correct for variation in long RNA-seq data are typically not well-suited for such small RNA-seq data, making it even more important to

Toward the human cellular microRNAome

“…Our data matched certain "ground truths" about specific microRNA expression in certain cell types, including miR-192 in epithelial cells, miR-126 in endothelial cells and platelets, and miR-9 in brain (McCall et al 2011a;Haider et al 2014;Kent et al 2014). As well, another recent paper investigating microRNA RNA-seq across multiple laboratories found relative quantification to be "remarkably accurate and reproducible" (Giraldez et al 2017), consistent with our work. Despite this, technical factors certainly drive some of the variation and clustering in these samples.…”

Large Differences in Small RNA Composition Between Human Biofluids

“…We used a uniform approach to RNA isolation, RNA-seq library preparation, sequencing library size selection, sequencing, and data analysis for all samples. The RNA-seq method we used reduces bias by employing oligonucleotides with random nucleotide sequences for adaptor ligation ( Giraldez et al, 2017 ). Use of a permissive size selection strategy rather than one targeting miRNAs allowed us to identify RNAs of many biotypes.…”

“…RNA was isolated from 200 μL of biofluid using the QIAGEN miRNEasy Micro Kit according to the manufacturer’s protocol, except that 1 mL Qiazol and 180 μL chloroform were used. Small RNA-seq libraries were prepared using 4N protocol D as previously described ( Giraldez et al, 2017 ; Supplemental Experimental Procedures ).…”

“…One objective of the NIH Extracellular RNA Communication Consortium (ERCC) is to identify the range of RNAs present in human biofluids. To address this goal, we compared the small RNA populations of a large and diverse collection of human biofluids using one standard RNA sequencing (RNA-seq) approach that was validated as part of a multicenter ERCC study ( Giraldez et al, 2017 ). This approach was designed for miRNAs but can also detect other small RNAs with a 5’ phosphate and a 3’ hydroxyl group.…”

xMD-miRNA-seq to generate near in vivo miRNA expression estimates in colon epithelial cells

“…Thus, a comparison across these methods may not be as robust as expected. 35 Since this project began on the premise of finding the most accurate in vivo cellular miRNA expression estimate for deconvoluting tissues, this will remain a challenge if tissue data is generated in a library format that is not fully compatible with the library preparation method of xMD-miRNA-seq. In fact, while the levels of miR-192/215 were 294,000 and 184,000 in two colon samples sequenced using the Illumina system (SRR837836, SRR837842) the RPM value for miR-192/215 was lower, at 44,000-70,000 in the xMD-Epi samples.…”