We present an update to the PIGEON (Phages and Integrated Genomes Encapsidated or Not) reference database of DNA viral sequences (vOTUs, mostly dsDNA bacteriophages) from global ecosystems. To reflect the inclusion of only virus size-fractionated metagenome- (virome-)derived vOTUs, we reintroduce the database as PEONY (Phages Encapsidated ONlY).
Knowledge of the global biogeography of DNA viruses, particularly dsDNA bacteriophages (phages, viruses of bacteria) is rapidly expanding. Opportunities to synthesize these efforts are critical to understanding fundamental viral ecology, phage-host dynamics, and links to global biogeochemical cycling. Existing reference datasets focused on phages comprise a mix of viral sequences predicted from total DNA metagenomes and RNA metatranscriptomes, isolated viral genomes, and viral size-fractionated metagenomes (viromes). The key advance here is the focus exclusively on viral sequences recovered from viromes, vastly reducing the potential for ‘false-positive’ viral sequences that likely exist in similar databases compiled from bioinformatic mining of a variety of multi-omic data types.
PIGEON1.0 was first assembled and made publicly available in 2021 with 266,125 viral operational taxonomic units (vOTUs > 10 kbp, clustered at 95% average nucleotide identity, ANI) from peat, soil, freshwater, marine, mammalian digestive, and plant-associated environments. The database was updated in 2023 (PIGEON2.0) to include additional vOTUs totaling 515,763. Due to the limited numbers of sequenced viromes at the time, particularly from soil, PIGEON1.0 and 2.0 included vOTUs recovered from both viromes and total DNA metagenomes. The total number of PIGEON vOTUs increased to 1,351,223 (PIGEON3.0). After removing 263,711 vOTUs from total metagenomes, 1,072,861 vOTUs remain in PEONY1.0. Subsequent minor updates have increased PEONY vOTUs to 1,088,256 (v1.3).
PEONY vOTUs encompass terrestrial, aquatic, and mammalian digestive ecosystems across five continents. Notably, 57% of vOTUs in PEONY originate from California, USA, where soils from various ecosystems (agricultural fields, grasslands, forests, wetlands, chaparral, and plant rhizospheres) have been routinely sampled by the Emerson Lab at the University of California, Davis and collaborators in recent years. This database provides a useful tool for studies of dsDNA viral (presumably, mostly bacteriophage) diversity and environmental distributions. Researchers can use PEONY to compare viruses from individual studies to global distribution patterns. For example, mapping vOTUs from fire-affected soils to the PIGEON2.0 database allowed Geonczy et al. to identify potentially fire-adapted vOTUs shared amongst other post-wildfire studies. As more viromics data becomes available we expect to update PEONY biennially.
The databases (both PIGEON3.0 and PEONY) and associated code for database visualization in R, scripts for curation, and additional methods information and associated metadata are available here.

Literature Search:
A Web of Science search was created on 12/01/25, using the search terms below, which yielded 3109 studies. The studies were split equally and randomly among four researchers (the first four authors of this resource announcement), using the R metagear package for Y/N/M (yes/no/maybe) screening for potential inclusion. The studies underwent a preliminary, manual title-abstract screening to eliminate total metagenomes, targeted sequencing, reviews, perspectives, spike-ins, and RNA-exclusive studies, resulting in 689 studies for full screening. 

Search terms: 
TOPIC: Virom OR “viral metagenom*” OR metavirome AND (“viral fraction” OR "0.22 µm" OR “0.2 µm” OR "0.45 µm" OR "0.1 µm" OR ultracentrifugation OR “iron chloride” OR ultracentrifuge OR TFF OR “tangential flow” OR PEG)   *

Date period: 2021-present 

The remaining 689 studies were re-randomly distributed amongst the same four researchers so that every manuscript was screened by two different researchers between the first and second screening efforts. The inclusion criteria are listed below: 

Studies were included if samples were size-fractionated viral metagenomes (< 0.22 µm or < 0.45 µm) and readily available at the assembled contig or clustered vOTU level. No reviews, perspectives, meta-analyses, mining from total metagenomes, exclusively RNA studies, multiple-displacement amplification (MDA), or dual RNA/DNA extractions followed by reverse transcription and combined sequencing were included. This screening resulted in a list of 16 studies that underwent bioinformatic processing. For datasets from which the data availability was unclear, the corresponding authors were emailed with a request for data access, which resulted in the inclusion of two additional studies. The full list of search results and subsequent screening are included on the project Github.

Inclusion of new and as-yet unpublished data from the Emerson Lab:
All published studies from the Emerson lab that met the laboratory and sample processing inclusion criteria above were included. In addition, unpublished viromic datasets with presumed final vOTU sequences were included, at the discretion of both Emerson and the current first author of each study. Further information is reported in the associated metadata file.

Processing: 
Viral contigs and vOTUs were re-processed for viral sequence prediction via geNomad v1.9.4 for consistency and filtered to > 10,000 bp in length. We then individually (per-study) dereplicated, aligned, and clustered the contigs into vOTUs with vclust v1.3.1 (leiden algorithm, ANI threshold 0.95, coverage threshold 0.85) prior to reclustering the full database(s).

Changes after Jul 3, 2026: Updated PEONY (now v1.1) to include one dataset that was erroneously removed when separating PIGEON3 and PEONY (Global Viromes, Gregory et al. 2019) and removed two within-lab datasets that contained mixed total metagenomes and viromes (headers "biochar_" and RR_2018"). The text was also updated to reflect these changes and report the new total number of vOTUs included in PEONY. Grant Gogul was added as a co-author.

Changes after Sep 10, 2026: Updated PEONY (now v1.3) to include one fixed duplicate header in Holmfeldt dataset (v1.2). Added Ignacio-Espinoza2020 dataset and reclustered, new total vOTUs = 1,088,256. Updated several headers for better connectivity to metadata. All changes to headers are recorded in the 'header_notes' column of the metadata file.