Persistent human DNA viruses are globally distributed, commonly acquired early in life, and leave genomic traces in host tissues, thus illustrating viral evolution and reflecting evolutionary relationships with human populations. This viral fingerprint may complement human genomic data as biogeographic markers, especially in contexts where reference data is unavailable or scattered.

We compiled 24,441 near‑full‑length genomes from NCBI GenBank, representing 29 human DNA viruses from five families (Polyomaviridae, Herpesviridae, Papillomaviridae, Parvoviridae, and Hepadnaviridae) with associated metadata obtained from database entries and primary literature. Phylogenetic analyses were performed using BEAST2, and machine learning models were implemented via the mGPS framework to predict the origin of viral sequences.

Phylogenetic reconstructions demonstrated continent‑ and country‑level clustering for many viruses, including BKPyV, JCPyV, HHV1, HHV2, HHV3, HHV6B, HCMV, HPV16, HPV18, HPV31, HPV45, and HBV. Geographic patterns were strongest in regions with adequate sequence representation, while global data availability remained highly uneven. Sixteen viruses met the sample‑size requirements for mGPS modelling, achieving weighted F1 test scores ranging from 0.7663 to 0.9980, indicating continent‑level predictivity despite dataset imbalance.

Several persistent human DNA viruses (14 out of the studied 29) show potential as biogeographic markers, offering complementary tools for forensic identification and population studies. Integrating phylogenetics and machine learning can enable host origin inference, though more globally even distribution of publicly available viral sequences, metadata standards, and ethical frameworks are essential for future applications.

Authors

• Martta Keskitalo (University of Helsinki, Finland)
• Matti Heino (University of Helsinki, Finland)
• Jianye Ge (Center for Human Identification and Department of Microbiology, Immunology, and Genetics, University of North Texas Health Science Center, United States of America)
• Bruce Budowle (University of Helsinki; Forensic Science Institute, Radford University; Othram Inc., United States of America)
• Mari Toppinen (University of Helsinki, Finland)
• Antti Sajantila (University of Helsinki; Finnish Institute for Health and Welfare, Finland)