A Genomic Language Model for Chimera Artifact Detection in Nanopore Direct RNA Sequencing

Area of Science:

• Genomics
• Bioinformatics
• Molecular Biology

Background:

• Chimera artifacts in nanopore direct RNA sequencing (dRNA-seq) can compromise transcriptome analysis accuracy.
• Current basecalling models struggle with the detection and removal of these artifacts.
• Adapter sequences contribute to chimeric read formation, complicating downstream analyses.

Purpose of the Study:

• To develop a novel method for precise identification and removal of adapter sequences from dRNA-seq reads.
• To eliminate chimeric read artifacts without relying on raw signal or alignment data.
• To improve the accuracy of downstream transcriptomics analyses using nanopore sequencing.

Main Methods:

• Introduction of DeepChopper, a genomic language model for artifact detection.
• Application of DeepChopper to base-called dRNA-seq long reads.
• Single-base resolution identification and removal of adapter sequences.

Main Results:

• DeepChopper effectively identifies and removes adapter sequences, eliminating chimeric artifacts.
• The method operates independently of raw signal or alignment information.
• Improved accuracy in downstream analyses such as transcript annotation and gene fusion detection.

Conclusions:

• DeepChopper significantly enhances the reliability of nanopore dRNA-seq data.
• The tool provides a robust solution for addressing chimera artifacts in RNA sequencing.
• DeepChopper advances the utility of nanopore dRNA-seq for comprehensive transcriptomics research.