Virgo Benchmarking Datasets

1. Global Ocean Eukaryotic Viral (GOEV) Database

Source: Extract from Gaïa M. et al. Nature (2023)
Components:

591 MAGs from Schulz, F. et al. (2020) [DOI: 10.1038/s41586-020-1957-x]

445 MAGs from Sunagawa, S. et al. (2020) [DOI: 10.1038/s41579-020-0364-5]

218 MAGs from Moniruzzaman, M. et al. (2020) [DOI: 10.1038/s41467-020-15507-2]

158 reference viral assemblies

Accessed: July 20, 2024
Data File: GOEV_DB_CONTIGS.db.zip from Figshare
Selection Criteria: Only contigs labeled at the Order taxonomic level were retained
Sampling Method: Not applicable
Final Sample Size: 1,412 viral contigs

--

2. Known Viral Sequence Clusters (kVSCs)

Source: Extract from Zolfo, M. et al. (2024) [DOI: 10.1101/2024.02.19.580813]
Data Files:

VSC5_rep_fnas_nr99_45k_metaphlanDB.fna.gz

VSCs_groups.csv metadata

Downloaded From: Zenodo, last accessed June 28, 2024
Selection Criteria:

Started from 45,872 representative sequences from MetaPhlan 4.1

Selected kVSCs (sequences clustering with a RefSeq representative)

Verified RefSeq accessions against ICTV Release #39 for accurate labeling

Sampling Method: RefSeq matching based on metadata
Final Sample Size: 2,232 representative sequences

--

3. ICTV Release #39

Source: International Committee on Taxonomy of Viruses (ICTV) Release #39
Downloaded Using: ICTVdump tool on July 17, 2024
Selection Criteria:

Viruses present in both VMR releases #37 and #39

At least two representatives per family

Sampling Method: Up to 5 genomes randomly sampled per family using pandas.sample(), 192 families represented.
Final Sample Size: 860.

--

4. RefSeq Viral Dataset (Random Iteration)

Source: NCBI Virus Portal NCBI Virus accessed on January 27, 2025
Selection Criteria:

Viruses with an assigned family-level taxonomy, up to 43 viruses per family

Sampling Method: Random uniform sampling
Final Sample Size: 6,778 viral genomes

--

5. RefSeq Viral Dataset (Prokaryote-Infecting)

Source: NCBI Virus Portal NCBI Virus accessed on January 27, 2025
Selection Criteria:

Viruses with an assigned family-level taxonomy

Prokaryote-infecting viruses only

Sampling Method: Random uniform sampling
Final Sample Size: 3,536 viral genomes

--

6. ICTV Release #39 (Reduction Study Subset)

Source: International Committee on Taxonomy of Viruses (ICTV) Release #39
Downloaded Using: ICTVdump tool on July 17, 2024
Selection Criteria: 1,000 viruses randomly sampled from the full release
Sampling Method: Random uniform sampling
Final Sample Size: 1,000 viral genomes

Notes: Reduction study starting data. We provide the source code for generating the fragmented genomes.