Sequencing accuracy and systematic errors of nanopore direct RNA sequencing.

Area of Science:

• Genomics
• Molecular Biology
• Bioinformatics

Background:

• Direct RNA sequencing (dRNA-seq) on Oxford Nanopore Technologies (ONT) platforms enables full-length transcript sequencing, including modifications and poly-A tail information.
• Despite its potential, the accuracy and error profiles of dRNA-seq remain insufficiently characterized.

Purpose of the Study:

• To comprehensively evaluate dRNA-seq accuracy and characterize systematic errors across diverse organisms and synthetic RNAs.
• To identify sequence contexts and signal-level features contributing to sequencing errors.
• To assess the impact of different basecallers and identify sources of data loss.

Main Methods:

• Analysis of dRNA-seq data from multiple species and synthetic RNAs using ONT kits SQK-RNA001 and SQK-RNA002.
• Systematic characterization of error types (deletions, insertions, mismatches) and their distribution.
• Examination of raw signal data to correlate errors with sequence context and signal features.
• Evaluation of basecaller performance and identification of adapter-related errors.
• Generation and analysis of dRNA-seq data using the SQK-RNA004 kit.

Main Results:

• Median read accuracy ranged from 87% to 92%, with deletions being the predominant error type.
• Heteropolymers and short homopolymers were major error contributors; cytosine/uracil-rich regions showed higher error rates.
• Systematic biases were observed at nucleotide and motif levels, linked to signal-level features.
• Read quality scores partially predict error rates; undetected DNA adapters contribute to errors.
• The newer SQK-RNA004 kit improved overall accuracy but retained similar systematic error patterns.

Conclusions:

• This study provides the first systematic analysis of dRNA-seq errors, revealing reproducible patterns.
• Identified signal-level insufficiencies and sequence context dependencies as sources of errors.
• Lays the groundwork for developing robust error correction methods for dRNA-seq data.