Reducing bias in RNA sequencing data: a novel approach to compute counts

Area of Science:

• Genomics
• Bioinformatics
• Molecular Biology

Background:

• Next-Generation Sequencing (NGS) is widely used for quantitative transcriptomics, offering an alternative to microarrays for gene expression analysis.
• Current RNA sequencing normalization methods rely on total read counts, which can be inaccurate due to non-uniform read distribution caused by sequencing errors and mapping ambiguities.
• This limitation affects the reliability of gene transcription level measurements.

Purpose of the Study:

• To introduce and evaluate a novel method, maxcounts, for quantifying exon expression in RNA sequencing.
• To compare the performance of maxcounts against the standard total read count approach.
• To assess robustness, accuracy, and independence from gene-specific covariates like exon length and GC-content.

Main Methods:

• Developed the maxcounts method, quantifying exon expression by the maximum of its per-base counts.
• Compared maxcounts with the standard total read count method using multiple RNA sequencing datasets.
• Evaluated performance based on independence from covariates, quantification accuracy, precision, and robustness to alignment quality variations.

Main Results:

• Both methods demonstrated high accuracy and low dependency on GC-content.
• Maxcounts exhibited less bias towards longer exons compared to the standard approach.
• Maxcounts showed reduced technical variability at low expression levels and enhanced robustness to alignment quality variations.

Conclusions:

• The standard read count method is sensitive to read distribution non-uniformity and feature length.
• Maxcounts provides a robust alternative, mitigating biases from non-uniform read distribution and exhibiting lower technical variability.
• Maxcounts is proposed as a superior approach for quantitative RNA sequencing applications.