Synthetic spike-in standards for RNA-seq experiments

Abstract

High-throughput sequencing of cDNA (RNA-seq) is a widely deployed transcriptome profiling and annotation technique, but questions about the performance of different protocols and platforms remain. We used a newly developed pool of 96 synthetic RNAs with various lengths, and GC content covering a 2 20 concentration range as spike-in controls to measure sensitivity, accuracy, and biases in RNA-seq experiments as well as to derive standard curves for quantifying the abundance of transcripts. We observed linearity between read density and RNA input over the entire detection range and excellent agreement between replicates, but we observed significantly larger imprecision than expected under pure Poisson sampling errors. We use the control RNAs to directly measure reproducible protocol-dependent biases due to GC content and transcript length as well as stereotypic heterogeneity in coverage across transcripts correlated with position relative to RNA termini and priming sequence bias. These effects lead to biased quantification for short transcripts and individual exons, which is a serious problem for measurements of isoform abundances, but that can partially be corrected using appropriate models of bias. By using the control RNAs, we derive limits for the discovery and detection of rare transcripts in RNA-seq experiments. By using data collected as part of the model organism and human Encyclopedia of DNA Elements projects (ENCODE and modENCODE), we demonstrate that external RNA controls are a useful resource for evaluating sensitivity and accuracy of RNA-seq experiments for transcriptome discovery and quantification. These quality metrics facilitate comparable analysis across different samples, protocols, and platforms.[Supplemental material is available for this article.]High-throughput sequencing applications are revolutionizing genome-wide analysis (Mardis 2008;Mortazavi et al. 2008;Celniker et al. 2009;Morozova et al. 2009;Gerstein et al. 2010;Metzker 2010;Roy et al. 2010). RNA-seq offers single-nucleotide resolution, strand specificity, and short-range connectivity through pairedend sequencing. Because of these strengths, there has been great interest in using RNA-seq to distinguish isoforms, calculate expression levels for transcripts, and uncover low abundance RNAs (He et al. 2008;Mortazavi et al. 2008;Nagalakshmi et al. 2008;Sultan et al. 2008;Wang et al. 2008Wang et al. , 2010Passalacqua et al. 2009;Gerstein et al. 2010;Roy et al. 2010;Trapnell et al. 2010;Berezikov et al. 2011;Graveley et al. 2011).While there are clear advantages to RNA-seq, it is less clear how well the procedure performs, as several studies have reported conflicting RNA-seq accuracy results. RNA-seq-determined concentrations of six in vitro synthetic transcripts show good linearity (Mortazavi et al. 2008), and in a study using quantitative PCR as the benchmark, RNA-seq showed better performance for genes with high expression, while two-channel microarrays were more sensitive in identifying differential expression between genes with low ex...

Improved definition of the mouse transcriptome via targeted RNA sequencing

“…Previous reports demonstrated that transcript models with at least eightfold sequence coverage can be confidently assembled (Jiang et al 2011). Thus, for each sample, we estimated the known concentration and the FPKM for which the ERCC controls reach a lower limit of 8× sequence coverage (Supplemental Figs.…”

“…Previous work (Jiang et al 2011) showed that 43% of sequencing reads aligned to the 1.5% most common transcripts, whereas only 1% of reads mapped to the 44% least abundant transcripts. This low depth of coverage inhibits current ab initio or de novo assemblers from identifying accurate transcript models (Steijger et al 2013), hampering evolutionary conservation analysis, transcript quantification, and differential expression analyses.…”

“…Transcript assemblies with less than eightfold coverage can be considered unreliable and discarded (Jiang et al 2011). To estimate an FPKM threshold corresponding to eightfold coverage, we calculated a second-degree polynomial fit between mean ERCC coverage [log 2 (read counts × read length/ERCC size)] and known concentrations (log 2 ) for each sample; based on this, we calculated the concentration for which the fit value equals 8 (Supplemental Fig.…”

Dual RNA-seq of pathogen and host

“…Importantly, strand-specificity should be guaranteed to be able to assess expression of anti-sense transcripts in both organisms. Moreover, as an internal quality control, biological samples should be supplemented with artificially synthesized RNA molecules (spike-in RNAs) (Jiang et al, 2011). The primary sequence of any such spike-in RNA must be confirmed bioinformatically to be absent from all the genomes under investigation.…”

“…Early time points were sequenced to an overall greater depth (~35 M reads/library) than the later ones (~25 M reads/library) to account for the expected lower fraction of Salmonella RNA in the early samples. (Jiang et al, 2011).…”