Zea mays RNA-seq estimated transcript abundances are strongly affected by read mapping bias

Area of Science:

• Genomics
• Transcriptomics
• Bioinformatics

Background:

• Genetic variation drives phenotypic diversity in species.
• RNA-sequencing (RNA-seq) commonly uses a single reference genome for read mapping.
• Dissimilar alleles may lead to mapping errors and underestimated transcript levels, especially in diverse species.

Purpose of the Study:

• Investigate RNA-seq mapping bias in maize (Zea mays).
• Determine if chromosomal features influence mapping bias.
• Assess the impact of mapping bias on gene expression quantification.

Main Methods:

• Utilized RNA-seq data from two distinct maize inbred lines (B73 and Mo17).
• Aligned RNA-seq reads to both B73 and Mo17 reference genomes.
• Analyzed gene expression estimates and their correlation with chromosomal locations.

Main Results:

• Significant mapping bias observed, with one line showing 2-4 times fewer positively acting alleles when mapped to the other's genome.
• Over half of detected alleles were missed when using a non-native reference genome.
• Bias was more pronounced at chromosomal ends and less in pericentromeric regions.
• Mapping bias affected untranslated regions more than splice junctions.
• Bias was consistent across different software and alignment parameters.

Conclusions:

• Mapping bias substantially impacts gene transcript abundance estimates in maize.
• Bias severity varies across different chromosomal features.
• Accurate transcript estimation in genetically variable maize individuals requires individual genome or transcriptome templates.