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

Sayed Mohammad Ebrahim Sahraeian, Marghoob Mohiyuddin, Robert Sebra, Hagen Tilgner, Pegah Tootoonchi Afshar, Kin Fai Au, Narges Bani Asadi, Mark Gerstein, Wing Hung Wong, M Snyder, Eric E. Schadt, Hugo Y. K. LamNature Communications 2017 · DOI

RNA-sequencing (RNA-seq) has revolutionized transcriptome studies, yet the proliferation of analysis tools has created a complex landscape for researchers. While many studies benchmark individual steps like alignment or quantification, few evaluate the entire end-to-end workflow. This study presents an exhaustive benchmark of RNA-seq analysis pipelines, evaluating 39 tools across approximately 490 combinations. The authors introduce RNACocktail, a high-accuracy, "broad-spectrum" pipeline that integrates short- and long-read data to provide deeper biological insights into transcript expression, variant calling, RNA editing, and fusion detection.

The Need for a Comprehensive Protocol

Standard RNA-seq protocols often focus narrowly on gene expression, potentially missing critical biological information. Furthermore, many established "gold standard" workflows, such as the original Tuxedo suite, have been surpassed by newer, more efficient tools. This research aims to provide a comprehensive guide that "unleashes the power" of RNA-seq by evaluating entire workflows across diverse biological contexts, including germline, cancer, and stem cell data.

Methods and the RNACocktail Pipeline

The researchers conducted a massive comparative study involving 39 analysis tools across four major categories: isoform detection/quantification, differential expression, variant calling/RNA editing, and fusion detection. A key feature of the study is the evaluation of hybrid sequencing, comparing traditional short-read (Illumina) data with long-read (PacBio) data. They specifically assessed hybrid tools like IDP (Isoform Detection and Prediction), which use short reads to correct and anchor long-read sequences.

The broad-spectrum RNA-seq analysis workflow, encompassing short-read and long-read isoform detection, variation analysis, and expression analysis.

The study utilized 15 samples across several well-characterized datasets, including NA12878 (human germline), MCF-7 (breast cancer), H1-hESC (embryonic stem cells), and SEQC (benchmark with known spike-ins).

Key Insights from the Benchmark

The benchmarking revealed several critical insights into modern transcriptome analysis:

  • The "Tuxedo" Protocol is Outdated: The traditional TopHat/Cufflinks workflow was consistently outperformed. HISAT2 was significantly faster and more precise for alignment, while StringTie excelled at transcript reconstruction.
  • Long Reads Resolve Complexity: Short reads struggle with highly polymorphic regions like the MHC. Long-read technologies and hybrid tools like IDP can span multiple exons, identifying crucial immune-related genes that short-read-only methods miss.
  • Alignment-Free Efficiency: For abundance estimation, alignment-free tools like Salmon and kallisto were faster and more consistent across samples than traditional alignment-based methods.
  • Distinguishing Variants from Edits: Most false positives in RNA variant calling are actually RNA editing events. Using a "genome-aware" approach or tools like GIREMI effectively distinguishes true genomic variants from post-transcriptional edits.

Performance Results

The proposed RNACocktail pipeline achieved higher accuracy in detecting biologically relevant differentially expressed genes compared to the standard Tuxedo protocol. Notably, using long reads via the IDP tool improved transcript-level precision by more than 20% over short-read-only assemblers.

Sensitivity and precision metrics at the gene and transcript levels across different datasets and tool combinations.

In terms of computational efficiency, HISAT2 was found to be 2.5x to 100x faster than STAR and TopHat, respectively, while maintaining the highest junction validation rate. For cancer research, FusionCatcher and SOAPfuse were the most sensitive short-read tools for fusion detection, while IDP-fusion (utilizing long reads) provided the highest precision.

Conclusion

While long-read sequencing provides superior structural insight, its higher cost and raw error rate necessitate hybrid correction methods. RNACocktail offers a robust, comprehensive solution for researchers seeking to maximize the biological information extracted from RNA-seq data. Future optimizations will likely target single-cell RNA-seq and the computational bottlenecks of long-read error correction.

Source: Sahraeian, S.M.E., et al. (2017). Gaining comprehensive biological insight into the transcriptome by performing a broad-spectrum RNA-seq analysis. Nature Communications. DOI: 10.1038/s41467-017-00050-4

Summary updated

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.RNA-seq is widely used for transcriptome analysis. Here, the authors analyse a wide spectrum of RNA-seq workflows and present a comprehensive analysis protocol named RNACocktail as well as a computational pipeline leveraging the widely used tools for accurate RNA-seq analysis.
BibTeX
@inproceedings{sahraeian2017gaining,
  author = {Sayed Mohammad Ebrahim Sahraeian and Marghoob Mohiyuddin and Robert Sebra and Hagen Tilgner and Pegah Tootoonchi Afshar and Kin Fai Au and Narges Bani Asadi and Mark Gerstein and Wing Hung Wong and M Snyder and Eric E. Schadt and Hugo Y. K. Lam},
  title = {{Gaining comprehensive biological insight into the transcriptome by performing a broad-spectrum RNA-seq analysis}},
  booktitle = {Nature Communications},
  year = {2017},
  doi = {10.1038/s41467-017-00050-4},
}
Figure