CoChat
Sign Up
Cited by 386Genomics and Phylogenetic StudiesSingle-cell and spatial transcriptomicsRNA Research and Splicing

The DOI matches its Crossref record; 76 of the 76 references it cites were found in a registry or an index.

Gaining comprehensive biological insight into the transcriptome by performing a broad-spectrum RNA-seq analysis

Open access
Sahraeian, Sayed Mohammad Ebrahim, Mohiyuddin, Marghoob, Sebra, Robert, Tilgner, Hagen, Afshar, Pegah T., Au, Kin Fai, Bani Asadi, Narges, Gerstein, Mark B., Wong, Wing Hung, Snyder, Michael P., Schadt, Eric, Lam, Hugo Y. K.·2017·Nature Communications
DOI 10.1038/s41467-017-00050-4PMC PMC5498581
Open this paper in CoChat — free accountOpen PDF

Abstract

RNA-sequencing (RNA-seq) is an essential technique for transcriptome studies, hundreds of analysis tools have been developed since it was debuted. Although recent efforts have attempted to assess the latest available tools, they have not evaluated the analysis workflows comprehensively to unleash the power within RNA-seq. Here we conduct an extensive study analysing a broad spectrum of RNA-seq workflows. Surpassing the expression analysis scope, our work also includes assessment of RNA variant-calling, RNA editing and RNA fusion detection techniques. Specifically, we examine both short- and long-read RNA-seq technologies, 39 analysis tools resulting in ~120 combinations, and ~490 analyses involving 15 samples with a variety of germline, cancer and stem cell data sets. We report the performance and propose a comprehensive RNA-seq analysis protocol, named RNACocktail, along with a computational pipeline achieving high accuracy. Validation on different samples reveals that our proposed protocol could help researchers extract more biologically relevant predictions by broad analysis of the transcriptome.

Overview

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1038/s41467-017-00050-4

Cites

76 of 76 references verified0 citation errors · 0 not checkable

Statistics

Figures

Fig. 1 The RNACocktail analysis protocol. RNACocktail is a comprehensive protocol of RNA-seq data analysis. The figure summarizes the widely used approaches for the key steps over the broad spectrum of RNA-seq analysis and also succinctly captures the possible workflows one can use to analyse RNA-seq data
Fig. 1 The RNACocktail analysis protocol. RNACocktail is a comprehensive protocol of RNA-seq data analysis. The figure summarizes the widely used approaches for the key steps over the broad spectrum of RNA-seq analysis and also succinctly captures the possible workflows one can use to analyse RNA-seq dataCC BY · PubMed Central
Fig. 2 Performance of different alignment schemes. a Overlap between the detected splice junctions by different schemes and their validation rate on reliable junctions in dbEST database74. A reliable EST junction set consists of junctions supported by at least two ESTs. The sizes of the circles reflect the number of junctions called by each scheme. For each tool, the number of junctions called and the validation rates (in parentheses) are shown. Validation rates for each subset of junctions are also shown on the Venn diagram. b Read mapping analysis: distribution of mapping status of sequenced fragments (left) (for NA12878, MCF7, and SEQC samples, mapping status for paired-end reads are shown, while for hESC, the distribution reflects percentage of uniquely mapped (blue), multi-mapped (orange), and unmapped (red) single-end reads), distribution of number of soft-clipped bases in mapped fragments (middle), distribution of the number of mismatches in mapped fragments (right)
Fig. 2 Performance of different alignment schemes. a Overlap between the detected splice junctions by different schemes and their validation rate on reliable junctions in dbEST database74. A reliable EST junction set consists of junctions supported by at least two ESTs. The sizes of the circles reflect the number of junctions called by each scheme. For each tool, the number of junctions called and the validation rates (in parentheses) are shown. Validation rates for each subset of junctions are also shown on the Venn diagram. b Read mapping analysis: distribution of mapping status of sequenced fragments (left) (for NA12878, MCF7, and SEQC samples, mapping status for paired-end reads are shown, while for hESC, the distribution reflects percentage of uniquely mapped (blue), multi-mapped (orange), and unmapped (red) single-end reads), distribution of number of soft-clipped bases in mapped fragments (middle), distribution of the number of mismatches in mapped fragments (right)CC BY · PubMed Central
Fig. 3 Performance of different transcriptome reconstruction schemes. a Distribution of number of exons per transcripts for different transcriptome reconstruction algorithms. Labels reflect the assembler, the long-read aligner (for IDP), and the short-read aligner used, respectively, with “-” separation. b Sensitivity and precision of different transcriptome reconstruction approaches at gene and transcript levels. The GENCODE reference transcriptome annotation is used as the truth set. The evaluations on a more recent update of MCF7 sample using the Iso-Seq pipeline resulted in a similar performance with only slight improvement. The union approaches that combined predictions from short reads and long reads (shown with a “+” in the label) slightly improved the performance of short-read isoform prediction schemes
Fig. 3 Performance of different transcriptome reconstruction schemes. a Distribution of number of exons per transcripts for different transcriptome reconstruction algorithms. Labels reflect the assembler, the long-read aligner (for IDP), and the short-read aligner used, respectively, with “-” separation. b Sensitivity and precision of different transcriptome reconstruction approaches at gene and transcript levels. The GENCODE reference transcriptome annotation is used as the truth set. The evaluations on a more recent update of MCF7 sample using the Iso-Seq pipeline resulted in a similar performance with only slight improvement. The union approaches that combined predictions from short reads and long reads (shown with a “+” in the label) slightly improved the performance of short-read isoform prediction schemesCC BY · PubMed Central
Fig. 4 Performance of different de novo transcriptome assembly techniques. a Distribution of transcript length. b N10-N50 values. (Nx is the contig length for which at least x% of the assembled transcript nucleotides were found in contigs that were at least of Nx length.) c ExN50 value at different expression percentiles. For ExN50, the N50 statistic was computed at percentile x, for only the top most highly expressed transcripts that represent x% of the total normalized expression data. Expression values were measured using eXpress
Fig. 4 Performance of different de novo transcriptome assembly techniques. a Distribution of transcript length. b N10-N50 values. (Nx is the contig length for which at least x% of the assembled transcript nucleotides were found in contigs that were at least of Nx length.) c ExN50 value at different expression percentiles. For ExN50, the N50 statistic was computed at percentile x, for only the top most highly expressed transcripts that represent x% of the total normalized expression data. Expression values were measured using eXpressCC BY · PubMed Central
Fig. 5 Performance of transcript abundance estimators. a Clustering of different schemes based on the Spearman rank correlation of their log expressions on NA12878. b Distribution of log2-fold change of expressions between MCF7-100 and MCF7-300 samples. For each method, dashed line represents the mean of the distribution and the dotted lines represents the quartiles. c Percentage of expression disagreement between MCF7-100 and MCF7-300 samples when low-expressed transcripts are discarded with different thresholds
Fig. 5 Performance of transcript abundance estimators. a Clustering of different schemes based on the Spearman rank correlation of their log expressions on NA12878. b Distribution of log2-fold change of expressions between MCF7-100 and MCF7-300 samples. For each method, dashed line represents the mean of the distribution and the dotted lines represents the quartiles. c Percentage of expression disagreement between MCF7-100 and MCF7-300 samples when low-expressed transcripts are discarded with different thresholdsCC BY · PubMed Central
Fig. 6 Performance of differential gene expressions analysis tools on SEQC-A vs. SEQC-B samples. a Spearman rank correlation, root-mean-score-deviation (RMSD), and AUC-30 scores for qPCR measured genes. Spearman rank correlation and RMSD scores are measured between the log2-fold change of the qRT-PCR and RNA-seq tools. AUC-30 score represents the area under the ROC curve up to the false positive rate of 30%. b ROC analysis of qRT-PCR measured genes (left) and ERCC (right) genes. For each differential analysis tool the plot reflects average performance when different alignment-based and alignment-free tools are used for abundance estimation and error bar shows the maximum and minimum variations. Results for each tool combination are shown in Supplementary Figs. 30 and 35
Fig. 6 Performance of differential gene expressions analysis tools on SEQC-A vs. SEQC-B samples. a Spearman rank correlation, root-mean-score-deviation (RMSD), and AUC-30 scores for qPCR measured genes. Spearman rank correlation and RMSD scores are measured between the log2-fold change of the qRT-PCR and RNA-seq tools. AUC-30 score represents the area under the ROC curve up to the false positive rate of 30%. b ROC analysis of qRT-PCR measured genes (left) and ERCC (right) genes. For each differential analysis tool the plot reflects average performance when different alignment-based and alignment-free tools are used for abundance estimation and error bar shows the maximum and minimum variations. Results for each tool combination are shown in Supplementary Figs. 30 and 35CC BY · PubMed Central
Fig. 7 Performance of different variant calling (a–c) RNA editing (d, e) and RNA fusion (f) detection approaches. a Accuracy of detecting NA12878 high-confidence calls in NIST gold standard. The analysis is restricted to the expressed exons identified by Cufflinks or StringTie. b Overlap between variants predicted by GATK and SAMtools. c Distribution of the predicted mismatch types by GATK that are missed in NIST HC calls (in StringTie’s expressed exons) in different genomic regions. d Distribution of RNA editing events detected in different genomic regions for NA12878. For multiple-samples scheme, final editing sites include the rare variants in NA12878 that are supported by at least 3 out of 12 short-read samples in our analysis. For pooled-samples scheme, final editing sites include the rare variants in NA12878 that are supported by at least 20 reads in the pooled alignment. e Measurement of RNA editing detection accuracy when some portion of SNPs are hidden from GIREMI. FDR represent proportion of predicted edits that are among high-confidence genomic variants in NA12878. f Performance of different RNA fusion detection schemes on MCF-7 sample
Fig. 7 Performance of different variant calling (a–c) RNA editing (d, e) and RNA fusion (f) detection approaches. a Accuracy of detecting NA12878 high-confidence calls in NIST gold standard. The analysis is restricted to the expressed exons identified by Cufflinks or StringTie. b Overlap between variants predicted by GATK and SAMtools. c Distribution of the predicted mismatch types by GATK that are missed in NIST HC calls (in StringTie’s expressed exons) in different genomic regions. d Distribution of RNA editing events detected in different genomic regions for NA12878. For multiple-samples scheme, final editing sites include the rare variants in NA12878 that are supported by at least 3 out of 12 short-read samples in our analysis. For pooled-samples scheme, final editing sites include the rare variants in NA12878 that are supported by at least 20 reads in the pooled alignment. e Measurement of RNA editing detection accuracy when some portion of SNPs are hidden from GIREMI. FDR represent proportion of predicted edits that are among high-confidence genomic variants in NA12878. f Performance of different RNA fusion detection schemes on MCF-7 sampleCC BY · PubMed Central
Fig. 8 The current RNACocktail computational pipeline. The pipeline is composed of high-accuracy tools in each step for general-purpose RNA-seq analysis
Fig. 8 The current RNACocktail computational pipeline. The pipeline is composed of high-accuracy tools in each step for general-purpose RNA-seq analysisCC BY · PubMed Central

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

Show all 76 references

Cited by

Sorted by citations

See all 386

Related

Metadata

Authors
Sahraeian, S. M. E., Mohiyuddin, M., Sebra, R., Tilgner, H., Afshar, P. T., Au, K. F., Bani Asadi, N., Gerstein, M. B., Wong, W. H., Snyder, M. P., Schadt, E., & Lam, H. Y. K.
Affiliations
  • Icahn School of Medicine at Mount Sinai (US)
  • Stanford Medicine (US)
  • Stanford University (US)
  • University of Iowa (US)
  • Yale University (US)
Corresponding author
Lam, Hugo Y. K.
ORCID
Venue
Nature Communications
Publisher
Springer Science and Business Media LLC
Published
2017-07-05
Volume
8
Issue
1
Pages
59-59
ISSN
2041-1723 (ISSN-L)
DOI
10.1038/s41467-017-00050-4
PMC
PMC5498581
OpenAlex
W2731748842
Cited by
376
References
76
Citations by year
2026: 29 · 2025: 48 · 2024: 64 · 2023: 41 · 2022: 49 · 2021: 60 · 2020: 28 · 2019: 31 · 2018: 20 · 2017: 5
FWCI
9.38
Type
journal article
Language
en
Open access
Licence
Locations
  • Nature Communications (publishedVersion, cc-by)
  • Europe PMC (PubMed Central) (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • RePEc: Research Papers in Economics (submittedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
Funders
  • National Institutes of Health
  • Genentech
  • National Human Genome Research Institute (R01HG008759, P50HG007735, R01HG007834)
Topics
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Genomics and Phylogenetic Studies
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Single-cell and spatial transcriptomics
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › RNA Research and Splicing
Keywords
gene expression profiling, RNA editing, computational pipeline, tumor transcriptome
MeSH terms
Base Sequence, Cell Line, Humans, Embryonic Stem Cells, Transcriptome

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review