Benchmarking of RNA-sequencing analysis workflows using whole-transcriptome RT-qPCR expression data.

Area of Science:

• Genomics
• Molecular Biology
• Bioinformatics

Background:

• RNA-sequencing (RNA-seq) is the standard for whole-transcriptome gene expression quantification.
• Numerous algorithms exist for deriving gene counts from RNA-seq reads.
• Independent benchmarking is crucial to assess the accuracy of different RNA-seq quantification methods.

Purpose of the Study:

• To independently benchmark five RNA-seq workflows for gene expression quantification accuracy.
• To compare RNA-seq derived gene expression with quantitative real-time PCR (qPCR) data.
• To identify potential biases or limitations in RNA-seq quantification methods.

Main Methods:

• Utilized RNA-seq data from MAQCA and MAQCB reference samples.
• Processed sequencing reads using five distinct workflows: Tophat-HTSeq, Tophat-Cufflinks, STAR-HTSeq, Kallisto, and Salmon.
• Compared gene expression measurements against validated qPCR data for all protein-coding genes.

Main Results:

• All tested RNA-seq workflows demonstrated high correlation with qPCR expression data.
• Approximately 85% of genes showed consistent expression fold changes between RNA-seq and qPCR when comparing MAQCA and MAQCB samples.
• Each method identified a small, specific set of genes with inconsistent expression measurements, often smaller, with fewer exons, and lower expression.

Conclusions:

• RNA-seq is a reliable method for gene expression quantification, but method-specific discrepancies exist for certain genes.
• A subset of genes, typically smaller and lowly expressed, may require careful validation when using RNA-seq.
• Reproducibility of inconsistent gene sets across independent datasets highlights potential systematic biases in quantification.