A practical guide to searching and interpreting OmniVira.

OmniVira is a literature-based database for tracking virus-host evidence related to zoonotic transmission, viral spillover, host range, reservoir hosts, and cross-species infection. It organizes information from biomedical literature into structured records that connect viruses, hosts, countries or regions, evidence types, and source publications.

OmniVira should be interpreted as an evidence-curation database, not as a real-time outbreak surveillance system, prevalence database, or formal risk-ranking platform. Each record reflects evidence reported in the source literature and should be interpreted together with the original publication.

OmniVira focuses on PubMed-indexed biomedical literature relevant to viral zoonoses, spillover, host range, interspecies transmission, reservoir identification, receptor usage, molecular adaptation, and related evolutionary evidence.

The main curated evidence comes from primary research articles. Reviews, news items, editorials, letters, guidelines, clinical trials, and other non-primary literature may be retrieved during automated searching, but they are screened and either excluded from evidence extraction or retained only for audit and review purposes. Excluded literature should be preserved with exclusion reasons so that decisions can be checked later.

A literature record represents one source publication, usually identified by a PMID. It contains article-level metadata such as title, journal, publication date, abstract, authors, and links to the original source.

An evidence record represents one structured finding extracted from a publication. One article may generate zero, one, or many evidence records. For example, a single paper may report several virus-host associations, multiple host species, different countries, or different types of evidence. Therefore, evidence counts are not the same as publication counts.

Evidence types describe what kind of support a publication provides. A single article may contain multiple evidence records and multiple evidence types.

Global Search is useful when you want to search across the whole database at once. It can accept virus names, virus abbreviations, host names, taxonomic terms, PMIDs, article titles, locations, or evidence-related keywords.

Results are grouped by record type, such as evidence records, viruses, hosts, and literature. This is the best starting point when you are unsure whether a term refers to a virus, host, publication, or evidence statement.

Evidence Search is used to find structured virus-host evidence directly. Users can search by keyword, PMID, evidence type, confidence level, publication year, virus, host, country, or location.

Each evidence result should be interpreted as a curated finding linked back to its source literature. Evidence records are the best entry point for questions such as which hosts have evidence for a virus, or what evidence supports a virus-host association.

Virus Search is used to find standardized virus entities. Users can search by virus name, abbreviation, family, genus, species, subtype, or related taxonomic terms.

Virus pages summarize standardized taxonomy, linked host records, associated evidence types, countries or regions, and supporting literature. For influenza viruses, subtype information such as H5N1 or H7N9 should be treated as an important searchable field.

Host Search is used to find standardized host entities. Users can search by scientific name, common name, NCBI taxonomy identifier, order, family, genus, species, or broader host group.

Host pages summarize linked viruses, evidence records, evidence types, countries or regions, and supporting literature. Host records are especially useful for reviewing evidence related to reservoir hosts, susceptible hosts, experimental hosts, and surveillance targets.

Literature Search is used to find source publications. Users can search by PMID, title, abstract terms, journal, publication year, or evidence-related keywords.

A literature record may contain one or more linked evidence records. Some retrieved publications may not have curated evidence if they were excluded, were not primary research, lacked relevant extractable evidence, or are still awaiting review.

OmniVira is designed to support a daily literature update workflow. The update process retrieves new PubMed records using the configured search strategy, screens candidate literature, preserves exclusion reasons, downloads and parses available records, extracts evidence, normalizes virus, host, and country entities, and stages low-confidence or newly detected entities for review.

Because normalization and evidence extraction may involve automated methods and human review, newly added records can be updated or corrected after initial ingestion. Database content should therefore be treated as actively curated rather than static.