G Eric Bastien

The paper uses metagenomic analysis and machine learning to study viral diversity and infection dynamics in Microcystis blooms, revealing broad host ranges and environmental adaptations.

The paper introduces VHIP, a machine learning model that predicts virus-host interactions from genomic data, achieving 87.8% accuracy in predicting interactions at the species level.

The paper benchmarks 63 viral identification rulesets on mock metagenomes and aquatic samples, finding that combining rules increases recall but also false positives, with some combinations achieving high accuracy.