Virus detection in high-throughput sequencing data without a reference genome of the host

Area of Science:

• Virology
• Bioinformatics
• Computational Biology

Background:

• Novel virus discovery relies on next-generation sequencing (NGS) and computational analysis.
• Separating host from viral sequence reads is challenging, especially without a host reference genome.
• Low viral read counts and lack of de novo assembly pose difficulties for existing pipelines.

Purpose of the Study:

• To present a novel, modular computational pipeline for discovering viruses in host samples.
• To address limitations of existing pipelines, particularly in the absence of a host reference genome.
• To introduce a decoy module for assessing false classification rates and improving discovery accuracy.

Main Methods:

• Development of a modular computational pipeline for virus discovery.
• Implementation of a method that does not require a host reference genome.
• Inclusion of a decoy module to evaluate false positive rates.
• Validation using simulated data and experimental samples with known viral content.

Main Results:

• The pipeline successfully identifies novel viruses in host samples, including those lacking reference genomes.
• A decoy module was introduced to assess and report error rates, enhancing result quality.
• Viruses with low read coverage were identified and visually reviewed.
• The pipeline reproduced findings from experimental samples.

Conclusions:

• The developed pipeline is effective for virus detection across diverse host species.
• The modular design allows integration into existing pipelines or standalone use.
• The novel decoy approach provides a crucial quality control for virus discovery pipelines.
• The pipeline offers a valuable tool for identifying novel viruses in genomic data.