False positives in trans-eQTL and co-expression analyses arising from RNA-sequencing alignment errors.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Next-generation sequencing (NGS) short reads can misalign due to sequence similarity in distinct genomic regions.
• Downstream consequences of NGS read misalignment, particularly in RNA-sequencing (RNA-Seq), are not fully understood.
• Misalignment can impact gene expression quantitative trait locus (eQTL) and co-expression analyses.

Purpose of the Study:

• To assess the impact of RNA-Seq read misalignment on false positive rates in eQTL and co-expression analyses.
• To investigate the extent to which sequence similarity causes false positives in trans-eQTL studies.
• To develop a method for quantifying and mitigating alignment errors in RNA-Seq association studies.

Main Methods:

• Assessed potential for incorrect alignment of RNA-Seq reads causing false positives in eQTL and co-expression analyses.
• Analyzed trans-eQTLs from human RNA-Seq studies, considering uniquely aligned reads.
• Quantified potential for cross-mapping between all pairs of annotated human genes.

Main Results:

• Trans-eQTLs are highly susceptible to false positives caused by alignment to similar genomic regions, even with unique read alignment.
• Over 75% of trans-eQTLs in a standard pipeline may result from alignment errors due to sequence similarity.
• Mapping errors can lead to misleading replication of associations across studies.
• Quantified cross-mapping potential across the human genome.

Conclusions:

• Sequence similarity-driven misalignment is a significant source of false positives in RNA-Seq association studies, particularly trans-eQTLs.
• Cross-mapping data provides a resource to filter or flag potential false positives in eQTL and co-expression analyses.
• Filtering based on cross-mapping significantly impacts the detection of associations, false discovery rate, functional enrichment, and replication in RNA-Seq studies.