Detection theory in identification of RNA-DNA sequence differences using RNA-sequencing.

Area of Science:

• Genomics
• Transcriptomics
• Bioinformatics

Background:

• High-throughput sequencing technologies enable single-nucleotide resolution analysis of the transcriptome.
• RNA editing, such as A-to-I and C-to-U deamination, modifies RNA sequences.
• The existence and validity of other RNA-DNA sequence differences (RDDs) are debated, with technical artifacts often implicated.

Purpose of the Study:

• To evaluate the performance of RNA-Sequencing (RNA-Seq) and aligners in accurately identifying RNA-DNA sequence differences (RDDs).
• To assess the impact of alignment artifacts and sequencing errors on RDD detection sensitivity and false discovery rates.
• To determine the reliability of RDD detection and the evidence for non-canonical RDDs in humans.

Main Methods:

• Generated simulated RNA-Seq datasets with known RDDs to assess detection performance.
• Evaluated the effect of alignment artifacts and sequencing errors on sensitivity and false discovery rates.
• Assessed the efficacy of various filters in identifying and removing false positive RDDs.

Main Results:

• RNA-Seq, with appropriate thresholds, can achieve low false negative and false discovery rates (<10%) for RDD detection, even with sequencing errors.
• Filters effectively discriminate between true and false positive RDDs.
• Analysis of a human lymphoblastoid cell line identified ~6,000 RDDs, predominantly A-to-G edits (likely ADAR-mediated); non-canonical RDDs were associated with poorer alignments, suggesting weak evidence for their widespread occurrence.

Conclusions:

• RNA-Seq is a powerful tool for genome-wide RDD surveying when appropriate thresholds and filters are employed.
• The majority of non-canonical RNA-DNA differences observed in human samples are likely technical artifacts, not true biological events.
• Evidence for widespread non-canonical RNA-DNA differences in humans is weak.