Bidirectional subsethood of shared marker profiles enables accurate virus classification

Area of Science:

• Virology
• Bioinformatics
• Computational Biology

Background:

• Viral metagenomic sequencing generates vast amounts of data.
• Official virus taxonomy undergoes frequent updates, posing challenges for existing classification tools.
• Keeping bioinformatic classification methods synchronized with evolving taxonomy is difficult.

Purpose of the Study:

• To develop a computational tool for accurate virus family prediction from metagenomic data.
• To ensure the tool's compatibility with any version of the official virus taxonomy.
• To provide a resource that aids in the classification of viruses.

Main Methods:

• Development of a novel viral sequence similarity metric.
• Implementation of the VIRGO computer program utilizing this metric.
• Validation of VIRGO's performance on metagenomic data.

Main Results:

• VIRGO accurately predicts virus families from metagenomic data.
• Achieved an F1 score exceeding 0.9 for virus family classification.
• Demonstrated compatibility with all versions of the official virus taxonomy.

Conclusions:

• VIRGO is a valuable resource for the virology community.
• The program's design facilitates easy integration of updated virus taxonomy releases.
• Highlights the need for bioinformatic methods that adapt to curated biological resources.