SinProVirP: a signature protein-based tool for accurate and rapid profiling of human gut virome

SinProVirP is a signature protein-based tool developed for genus-level profiling of the human gut virome. To evaluate the performance of viral identification and quantification at the viral VC-level across different tools, we benchmarked SinProVirP, WGVP, and Phanta using simulated metagenomic datasets.

Accuracy dataset: Each dataset consists of 10 million 150-bp reads derived from 100 viral and 100 bacterial genomes, with exponentially decreasing relative abundance and a 9:1 bacterial-to-viral reads ratio. To assess tool performance on datasets with varying viral proportions, we also simulated datasets with 1%, 2%, 5%, 10%, 20%, and 30% viral reads, each with ten replications.

Comprehensiveness dataset: To detect both within-database (within-DB) species and novel species, we aligned 8,424 assembled viral genomes from the human gut, sourced from Lou et al., against the SinProVirP and Phanta databases. This resulted in the identification of 2,361 within-DB species and 198 novel species. We then generated ten simulated datasets for both novel and within-DB species, each containing 150-bp paired-end reads at ten-fold depth from 100 randomly selected species, generated using insilicoseq.