[Paper Review] RNA-Seq Mapping Errors When Using Incomplete Reference Transcriptomes of Vertebrates

Alexis Black Pyrkosz, Hans H. Cheng+1 more|arXiv (Cornell University)|Mar 11, 2013

Genomics and Phylogenetic Studies40 references14 citations

TL;DR

This study investigates how incomplete reference transcriptomes and alternative splicing introduce errors in RNA-Seq read mapping and expression quantification. Using simulated transcriptomes and reads, it demonstrates that missing transcripts reduce true positive mapping rates, while shared exons between splice variants increase false positives; grouping transcripts by gene or read-sharing improves accuracy, but only complete reference transcriptomes can fully resolve the issue, with longer reads (>1 kb) needed to reduce ambiguity.

ABSTRACT

Whole transcriptome sequencing is increasingly being used as a functional genomics tool to study non- model organisms. However, when the reference transcriptome used to calculate differential expression is incomplete, significant error in the inferred expression levels can result. In this study, we use simulated reads generated from real transcriptomes to determine the accuracy of read mapping, and measure the error resulting from using an incomplete transcriptome. We show that the two primary sources of count- ing error are 1) alternative splice variants that share reads and 2) missing transcripts from the reference. Alternative splice variants increase the false positive rate of mapping while incomplete reference tran- scriptomes decrease the true positive rate, leading to inaccurate transcript expression levels. Grouping transcripts by gene or read sharing (similar to mapping to a reference genome) significantly decreases false positives, but only by improving the reference transcriptome itself can the missing transcript problem be addressed. We also demonstrate that employing different mapping software does not yield substantial increases in accuracy on simulated data. Finally, we show that read lengths or insert sizes must increase past 1kb to resolve mapping ambiguity.

Motivation & Objective

• To assess the impact of incomplete reference transcriptomes on RNA-Seq read mapping accuracy and expression quantification.
• To evaluate how alternative splice variants that share exons contribute to mapping errors and false positive rates.
• To determine whether mapping software choice or read length/insert size affects mapping accuracy in incomplete transcriptomes.
• To investigate whether grouping transcripts by gene or read-sharing relationships reduces mapping errors.
• To evaluate the potential of longer reads (>1 kb) to resolve mapping ambiguity caused by shared exons or incomplete references.

Proposed method

• Simulated transcriptomes with random sequences (100–5000 bp) and alternative splice variants using five operations: truncation, extension, insertion, deletion, and substitution.
• Generated single- and paired-end reads (100 bp) from these transcriptomes with 1% random substitution errors and coverage levels of 0x, 10x, 100x, or 1000x.
• Mapped reads to the reference using Bowtie with default parameters, including unique (-m 1) and multimap (-a) modes.
• Assessed mapping accuracy by comparing the true transcript origin of each read to the transcript assigned by the mapper (true positive vs. false positive).
• Measured transcript expression accuracy by comparing actual and mapped read counts per transcript, flagging differences >2-fold as erroneous.
• Grouped transcripts by gene (using Ensembl relationships) and by multimapping read sharing to assess gene-level and transcript-family-level expression accuracy.

Experimental results

Research questions

• RQ1 How does an incomplete reference transcriptome affect the true positive rate of RNA-Seq read mapping?
• RQ2 To what extent do shared exons between alternative splice variants increase false positive mapping rates?
• RQ3 Can grouping transcripts by gene or read-sharing relationships reduce mapping errors in incomplete transcriptomes?
• RQ4 Does using different mapping software significantly improve mapping accuracy on simulated incomplete transcriptomes?
• RQ5 What read length or insert size is required to resolve mapping ambiguity in complex transcriptomes?

Key findings

• Incomplete reference transcriptomes significantly reduce the true positive mapping rate due to missing transcripts, leading to underestimation of transcript expression.
• Alternative splice variants that share exons increase the false positive rate, as reads map equally well to multiple transcripts, causing misassignment.
• Grouping transcripts by gene or read-sharing relationships reduces false positives, but cannot fully compensate for missing transcripts.
• Mapping software choice (e.g., Bowtie) does not yield substantial improvements in accuracy on simulated data, indicating that software alone cannot resolve mapping errors.
• Read lengths or insert sizes must exceed 1 kb to meaningfully reduce mapping ambiguity caused by shared exons or incomplete references.
• Accurate isoform-level expression quantification remains challenging; gene-level or transcript-family-level analysis is more robust but still limited by reference completeness.