Help & Documentation

Understand the evidence. Verify every claim.

OmniVira transforms fragmented literature and outbreak reports into traceable records for viral spillover, zoonotic transmission, host range, and emergence research. This guide explains what the database contains, how records are produced, and how they should be cited and interpreted.

Evidence
8,118
Virus
604
60 families
Host
777
55 orders
Source articles
2,633
Evidence ontology
v2.1.0
5 categories · 14 types
Data updated
12 August 2026
Daily workflow
01

Three-minute orientation

Start here

OmniVira is a curated viral spillover and zoonotic evidence knowledgebase. Its principal unit is a structured, source-linked evidence record rather than a publication alone or a binary virus-host association.

What OmniVira provides

Searchable literature, OVE evidence records, standardized virus and host entities, geographic annotations, WHO Disease Outbreak News monitoring, and provenance back to the original source.

What OmniVira does not provide

A clinical decision tool, a formal pathogen-risk score, a claim that every detected host is a reservoir, or a substitute for reading the original publication or official outbreak report.

Designed for Virologists Zoonotic disease researchers Veterinary and wildlife health Epidemiology and surveillance Bioinformatics and data science Public-health preparedness
02

How the records relate

From a publication to traceable evidence

During curation, one PubMed publication may generate zero, one, or multiple candidate evidence records. Public source-article counts include only publications linked to at least one publicly retrievable OVE record.

Source layer Indexed PubMed record Counted as a source article only when linked to at least one public OVE record
Evidence layer 0–N OVE records Category, type, key finding, supporting text, and source linkage
Entity layer Standardized links Virus, host, host context, country, raw mentions

Worked example

How to read OVE7661

Open record
Source
PMID 38140677
Evidence type
Molecular Adaptation
Key finding
SARS-CoV-2 genomes from mink and farm workers carried F486L and N501T mutations associated with adaptation in mink.
Traceability
The OVE record links the structured finding to its supporting text, standardized entities, source article, and extraction metadata.
03

Public methods overview

How OmniVira is constructed

1 Systematic retrieval Configured PubMed search and citation-based supplementation
2 Deduplication & quality filtering Duplicate, retracted, non-original, and non-extractable records are screened
3 LLM-assisted relevance triage Candidate literature is assessed for zoonotic and spillover relevance
4 Ontology-guided extraction Evidence type assignment precedes type-specific structured extraction
5 Entity normalization Virus, host, and country mentions are resolved against controlled authorities
6 Human-supervised curation Review decisions, corrections, exclusions, and provenance are retained
Literature inclusion and exclusion rules

Primary research with extractable evidence relevant to viral detection, isolation, serology, reservoirs, transmission, host range, experimental infection, receptor usage, molecular adaptation, and genomic evolution is prioritized.

Retracted records; errata, corrections, and corrigenda; news, comments, editorials, and letters; datasets and audiovisual records; non-original publication types; records with a missing title; and records without an abstract are excluded from V2 evidence extraction. Excluded candidates and their reasons are retained in the screening history for later audit rather than silently discarded.

Why citation-based supplementation is separate from daily retrieval

Citation-based supplementation is used to identify potentially important earlier studies that may not be recovered by a single keyword strategy. Because highly cited records contain more reviews, methods papers, software, and other out-of-scope material, they undergo stricter relevance screening before evidence extraction.

04

Controlled Evidence Ontology v2.1.0

Five categories and fourteen evidence types

A category is a high-level scientific domain. An evidence type is the controlled label assigned to an individual OVE record. Category names are never used as substitutes for a specific evidence type.

Evidence category

Zoonotic Surveillance

4 evidence types

Evidence from surveillance, sampling, virus detection, virus isolation, serology, or reservoir-ecology studies that documents viral presence, exposure, or ecological maintenance in hosts or host-associated environments relevant to zoonotic and spillover research.

Scope. This category captures observation and surveillance evidence. It does not by itself prove a transmission event unless source-recipient movement is explicitly supported.

Virus Isolation Evidence that infectious virus was recovered, cultured, or isolated from a host, sample, tissue, vector, environmental material, or host-associated source.

Definition

Evidence that infectious virus was recovered, cultured, or isolated from a host, sample, tissue, vector, environmental material, or host-associated source.

Boundary and exclusion rule

Use only when recovery/culture/isolation of infectious virus or isolate is supported; otherwise use Viral Detection.

Required type-specific fields

  • virus_name_raw Raw name of the virus, viral species, strain, subtype, or viral group recovered or cultured from the sample.
  • source_host_or_sample_raw Raw name of the host species, host group, vector, environmental source, specimen source, or animal-human interface source from which infectious virus was isolated.
  • isolation_outcome Reported outcome of virus isolation, including successful isolation, culture positivity, recovery of infectious virus, cytopathic effect, plaque formation, passage result, or comparable isolation result.

Recommended fields when explicitly supported

sample_type_raw, isolation_system, isolate_name_raw

Viral Detection Evidence that viral nucleic acid, antigen, viral material, or viral signal was detected in hosts, samples, vectors, or environments without necessarily demonstrating infectious-virus isolation.

Definition

Evidence that viral nucleic acid, antigen, viral material, or viral signal was detected in hosts, samples, vectors, or environments without necessarily demonstrating infectious-virus isolation.

Boundary and exclusion rule

Use for detection or surveillance signals without evidence of successful infectious-virus isolation.

Required type-specific fields

  • target_host_raw Raw name of the monitored or sampled animal host, host group, vector, environmental source, or animal-human interface source.
  • virus_name_raw Raw name of the virus, viral species, strain, subtype, sequence group, or viral group detected, discovered, or analyzed in surveillance specimens.
  • methods Methods used to monitor, detect, identify, sequence, or analyze viral material in surveillance specimens.

Recommended fields when explicitly supported

geographic_location_raw, sample_types, sample_size_raw

Serological Evidence Evidence of antibody, neutralization, seropositivity, seroconversion, or related immune exposure to a virus in a host population or sampled group.

Definition

Evidence of antibody, neutralization, seropositivity, seroconversion, or related immune exposure to a virus in a host population or sampled group.

Boundary and exclusion rule

Use for exposure evidence based on antibodies or neutralization; do not treat serology alone as proof of active infection or transmission.

Required type-specific fields

  • virus_name_raw Raw name of the virus, viral species, strain, subtype, antigen, or viral group targeted by serological or antibody evidence.
  • sampled_host_raw Raw name of the host species, host group, exposed population, or human/animal population from which serum or antibody evidence was obtained.
  • methods Serological, antibody, neutralization, or immunological method used to detect exposure or infection history.
  • serological_result_raw Raw text reporting seroprevalence, antibody positivity, seropositive proportion, neutralizing antibody result, titer, seroconversion, or antibody detection outcome.

Recommended fields when explicitly supported

sample_types

Reservoir Ecology Evidence characterizing ecological reservoirs, maintenance hosts, host populations, vectors, or ecological conditions associated with viral persistence, circulation, or exposure.

Definition

Evidence characterizing ecological reservoirs, maintenance hosts, host populations, vectors, or ecological conditions associated with viral persistence, circulation, or exposure.

Boundary and exclusion rule

Use for ecological reservoir or maintenance-host evidence; use Viral Detection for simple positive sampling without reservoir interpretation.

Required type-specific fields

  • virus_name_raw Raw name of the virus, viral species, strain, subtype, or viral group involved in ecological, reservoir-host, maintenance, or natural-circulation evidence.
  • target_host_raw Raw name of the target host, proposed reservoir host, maintenance host, host population, host community, vector, or ecological source studied for viral persistence or circulation.
  • methods Methods used for ecological, reservoir-host, maintenance, or natural-circulation analysis.

Recommended fields when explicitly supported

geographic_location_raw, sample_types, ecological_metrics

Evidence category

Transmission Evidence

3 evidence types

Evidence that describes, infers, or investigates viral movement between source and recipient hosts, including spillover, outbreak-associated transmission, and cross-species transmission among non-human hosts.

Scope. Use this category when the evidence concerns a transmission relationship or event, not merely detection of the same virus in multiple hosts.

Spillover Event Evidence describing a naturally occurring or epidemiologically supported animal-to-human viral transmission event in which a non-human animal source, reservoir, vector, or animal-derived exposure is linked directly to infection in a human individual or population.

Definition

Evidence describing a naturally occurring or epidemiologically supported animal-to-human viral transmission event in which a non-human animal source, reservoir, vector, or animal-derived exposure is linked directly to infection in a human individual or population.

Boundary and exclusion rule

Use only for animal-to-human transmission with a non-human animal source and a human recipient. Human-to-animal spillback and natural transmission between different non-human animal species belong to Cross-species Transmission; controlled transmission experiments belong to Experimental Transmission. Co-detection, serology, phylogenetic relatedness, ecological association, or speculative spillover risk alone is insufficient.

Required type-specific fields

  • virus_name_raw Raw name of the virus, viral species, strain, subtype, or viral group involved in the animal-to-human transmission event.
  • source_host_raw Raw name of the non-human animal source, reservoir, intermediate host, vector, or animal-derived exposure source in the animal-to-human transmission event.
  • recipient_host_raw Raw name of the human recipient, case, exposed occupational group, or affected human population in the animal-to-human transmission event.
  • transmission_direction Direction of the animal-to-human transmission event.

Recommended fields when explicitly supported

geographic_location_raw, study_design, methods

Outbreak Investigation Evidence from an outbreak, epidemic, cluster, or case investigation that links a virus to affected hosts, settings, source hypotheses, transmission context, or outbreak scale.

Definition

Evidence from an outbreak, epidemic, cluster, or case investigation that links a virus to affected hosts, settings, source hypotheses, transmission context, or outbreak scale.

Boundary and exclusion rule

Use for outbreak/case-cluster investigations; use Spillover Event only when animal-human transmission is specifically supported.

Required type-specific fields

  • virus_name_raw Raw name of the virus, viral pathogen, virus group, strain, subtype, or variant involved in the outbreak, epidemic, cluster, or case investigation.
  • outbreak_host_raw Raw name of the host species, host group, population, or human/animal group in which the outbreak occurred or was investigated.

Recommended fields when explicitly supported

suspected_source_host_raw, affected_host_raw, transmission_direction, outbreak_setting_raw, geographic_location_raw, outbreak_time_raw, outbreak_scale_raw, methods

Cross-species Transmission Evidence of naturally occurring or epidemiologically supported viral transmission from humans to non-human animals, or between different non-human animal species, including reverse zoonosis and directionally unresolved but well-supported cross-species host-pair transmission.

Definition

Evidence of naturally occurring or epidemiologically supported viral transmission from humans to non-human animals, or between different non-human animal species, including reverse zoonosis and directionally unresolved but well-supported cross-species host-pair transmission.

Boundary and exclusion rule

Use for human-to-animal spillback, transmission between different non-human animal species, or a well-supported interspecies host pair for which direction cannot be resolved. Animal-to-human transmission belongs to Spillover Event; controlled transmission experiments belong to Experimental Transmission. Same-species, human-to-human, susceptibility-only, receptor-only, co-detection, serology-only, phylogeny-only, or speculative host-jump statements are excluded.

Required type-specific fields

  • virus_name_raw Raw name of the virus, strain, isolate, subtype, variant, or viral group involved in human-to-animal spillback or transmission between different non-human animal species.
  • source_host_raw Raw name of the human or non-human source host, donor population, reservoir, or source population from which cross-species transmission originated or was inferred.
  • recipient_host_raw Raw name of the non-human animal recipient host, spillback host, host species, host group, or target population receiving or participating in the cross-species transmission.
  • transmission_direction Direction or host-pair relationship of a human-to-animal or non-human animal cross-species transmission event.

Recommended fields when explicitly supported

geographic_location_raw, study_design, methods

Evidence category

Experimental Infection

3 evidence types

Controlled experimental evidence evaluating host susceptibility, host range, pathogenicity, or transmission using defined viruses, hosts, model systems, challenge systems, or exposure designs.

Scope. Use this category for deliberate experimental systems rather than natural outbreaks or field surveillance.

Pathogenicity Experiment Controlled experimental evidence evaluating disease phenotype, pathogenicity, virulence, tissue damage, clinical outcome, or severity caused by a virus in a tested host system.

Definition

Controlled experimental evidence evaluating disease phenotype, pathogenicity, virulence, tissue damage, clinical outcome, or severity caused by a virus in a tested host system.

Boundary and exclusion rule

Use for controlled disease phenotype or virulence assessment; use Host Range Experiment when the question is host susceptibility rather than disease severity.

Required type-specific fields

  • virus_name_raw Raw name of the virus, strain, isolate, subtype, variant, mutant, or viral group used in the pathogenicity experiment.
  • tested_host_raw Raw name of the experimentally infected, challenged, exposed, or tested host used to evaluate pathogenicity.
  • experimental_system Experimental system used to evaluate pathogenicity, such as an animal model, humanized animal model, organoid, ex vivo tissue system, or other controlled infection model.
  • pathogenicity_outcomes Reported pathogenicity-related outcomes, including clinical signs, disease phenotype, weight loss, morbidity, mortality, lesions, pathology, tissue damage, viral shedding, host response, or attenuation/severity findings.
  • methods Experimental, virological, pathological, clinical-scoring, viral-load, histological, imaging, immunological, or analytical methods used to evaluate pathogenicity.
Host Range Experiment Controlled experimental evidence testing whether a virus can infect, replicate in, enter, or otherwise establish susceptibility across one or more host species or host-derived systems.

Definition

Controlled experimental evidence testing whether a virus can infect, replicate in, enter, or otherwise establish susceptibility across one or more host species or host-derived systems.

Boundary and exclusion rule

Use for controlled susceptibility/host-range testing; use Experimental Transmission when source-to-recipient spread is tested.

Required type-specific fields

  • virus_name_raw Raw name of the virus, strain, isolate, subtype, variant, mutant, or viral group used in the host-range experiment.
  • tested_host_raw Raw name of the host species, host group, animal model, cell-derived host context, organoid, tissue, or host system tested for susceptibility, permissiveness, infection, replication, or host range.
  • experimental_system Experimental system used to test host range, such as an animal challenge model, cell-culture system, organoid, ex vivo tissue, pseudovirus system, receptor-expression system, or other controlled host-range assay.
  • methods Experimental, virological, cell-culture, animal-challenge, receptor-entry, replication, pathology, sequencing, or analytical methods used in the host-range experiment.

Recommended fields when explicitly supported

virus_source_host_raw, sample_types

Experimental Transmission Controlled experimental evidence testing transmission from a source host or source system to a recipient host or recipient system under defined exposure conditions.

Definition

Controlled experimental evidence testing transmission from a source host or source system to a recipient host or recipient system under defined exposure conditions.

Boundary and exclusion rule

Use only when a source-to-recipient transmission design is present under controlled experimental conditions.

Required type-specific fields

  • virus_name_raw Raw name of the virus, strain, isolate, subtype, variant, mutant, or viral group used in the experimental transmission study.
  • source_host_raw Raw name of the donor, infected source, shedding source, inoculated source, vector source, or starting host in the experimental transmission setup.
  • recipient_host_raw Raw name of the recipient, contact, exposed, sentinel, vector-exposed, or target host tested for acquisition of infection in the experimental transmission setup.
  • experimental_system Controlled experimental system used to test transmission, such as animal transmission model, contact transmission model, aerosol model, vector transmission model, environmental exposure model, or co-housing model.
  • transmission_outcomes Reported transmission outcome categories or findings, such as recipient infection, contact transmission, aerosol transmission, vector-borne transmission, seroconversion, shedding in recipients, inefficient transmission, or no transmission.
  • methods Experimental, virological, serological, sequencing, epidemiological-model, animal-model, vector-assay, sampling, or analytical methods used to evaluate experimental transmission.

Evidence category

Functional Mechanism

2 evidence types

Mechanistic evidence describing molecular or cellular determinants of viral host range, entry, receptor usage, tissue tropism, adaptation, replication, immune escape, or other functional processes relevant to spillover potential.

Scope. Use this category when the evidence explains mechanism. Do not use it for genomic association alone without a functional interpretation.

Receptor Usage Functional evidence identifying or testing viral receptor usage, receptor binding, entry factors, or host-cell entry determinants.

Definition

Functional evidence identifying or testing viral receptor usage, receptor binding, entry factors, or host-cell entry determinants.

Boundary and exclusion rule

Use for receptor or entry-factor evidence; use Molecular Adaptation when the main claim is adaptive mechanism rather than receptor identity alone.

Required type-specific fields

  • virus_name_raw Raw name of the virus, viral species, strain, subtype, variant, pseudovirus, or viral system involved in receptor-usage or cell-entry evidence.
  • receptor_raw Raw name of the receptor, co-receptor, attachment factor, or receptor-like molecule explicitly implicated in viral binding, entry, tropism, or host-range compatibility.
  • methods Methods used to support receptor usage, receptor compatibility, viral binding, cell entry, or receptor-mediated host range.

Recommended fields when explicitly supported

host_factors, tested_host_raw

Molecular Adaptation Evidence linking viral or host molecular features to adaptation, host range, receptor binding, replication, immune escape, tropism, virulence, or transmission fitness.

Definition

Evidence linking viral or host molecular features to adaptation, host range, receptor binding, replication, immune escape, tropism, virulence, or transmission fitness.

Boundary and exclusion rule

Use for adaptive molecular mechanisms; do not assign this type to neutral genomic variation without mechanistic relevance.

Required type-specific fields

  • virus_name_raw Raw name of the virus, viral strain, isolate, subtype, variant, mutant, lineage, or viral group involved in molecular adaptation evidence.
  • mechanism_types Molecular adaptation mechanism category supported by the evidence, such as receptor binding, receptor usage, host entry, replication adaptation, immune escape, tropism, virulence adaptation, host-range expansion, or transmission fitness.

Recommended fields when explicitly supported

genes_or_proteins, mutations, receptors, host_factors

Evidence category

Genomic Evolution

2 evidence types

Genomic, phylogenetic, evolutionary, recombination, or reassortment evidence that informs viral emergence, host switching, lineage movement, adaptation history, or generation of novel viral genotypes.

Scope. Use this category for sequence-based evolutionary evidence. Functional receptor or pathogenicity assays should be assigned to their mechanism or experimental categories.

Recombination or Reassortment Evidence that viral recombination or genome-segment reassortment generated or contributed to a viral genotype relevant to emergence, host range, or transmission.

Definition

Evidence that viral recombination or genome-segment reassortment generated or contributed to a viral genotype relevant to emergence, host range, or transmission.

Boundary and exclusion rule

Use when recombination or reassortment is the evidence focus, not merely when a genome sequence is reported.

Required type-specific fields

  • recombinant_or_reassortant_virus_name_raw Raw name of the recombinant, reassortant, mosaic, segment-exchange, or mixed-ancestry virus, strain, subtype, lineage, or viral group described in the evidence record.
  • event_type Type of genomic exchange or mosaic event described by the evidence record.

Recommended fields when explicitly supported

parental_virus_a_raw, parental_virus_b_raw, parental_virus_names_raw, genes_or_segments

Phylogenetic Evolution Phylogenetic or genomic-evolution evidence describing viral lineage relationships, host-associated evolution, divergence, adaptation, or evolutionary origin.

Definition

Phylogenetic or genomic-evolution evidence describing viral lineage relationships, host-associated evolution, divergence, adaptation, or evolutionary origin.

Boundary and exclusion rule

Use for phylogenetic/evolutionary inference; use Recombination or Reassortment when genome exchange is the central claim.

Required type-specific fields

  • virus_name_raw Raw name of the virus, viral species, strain, subtype, lineage, clade, genotype, or viral group involved in phylogenetic or genomic-evolution analysis.
  • analysis_methods Evolutionary, phylogenetic, phylodynamic, phylogeographic, comparative genomic, lineage, molecular-clock, or related sequence-analysis methods used to support the evidence.

Recommended fields when explicitly supported

host_raw, genes_or_proteins

High-value boundary decisions

Animal → human, natural event Spillover Event
Human → animal or animal → different animal species Cross-species Transmission
Controlled donor-to-recipient experiment Experimental Transmission
Susceptibility, entry, or replication across host systems Host Range Experiment
Receptor identity, binding, or entry-factor evidence Receptor Usage
05

Transparent provenance

AI assistance and human-supervised curation

OmniVira uses AI to make evidence-scale curation feasible, but AI output, authority-based normalization, and curator approval are distinct states. The current beta contains active records at different stages of record-level review.

  1. AI-assisted Screening and candidate extraction

    LLMs support relevance triage, Evidence Type assignment, and initial field extraction.

  2. Validated Schema and boundary checks

    Required fields, controlled values, direction rules, duplicates, and source hashes are checked.

  3. Normalized Authority and mapping resolution

    Raw virus, host, and geographic mentions are mapped without discarding their original form.

  4. Human-supervised Correction, approval, or exclusion

    The Curation Console preserves reviewer decisions, reasons, changes, and audit history.

Pending review Active beta record awaiting final record-level curator decision.
Curator-reviewed Record inspected and accepted or corrected by an authorized curator.
Needs review Ambiguity or conflict requires additional expert assessment or full text.
Excluded Record is retained in the audit history but suppressed from public evidence results.
How AI-generated summaries differ from evidence records

Evidence-grounded Research Briefs are AI-assisted orientation summaries. They synthesize linked, publicly active OVE records as the factual basis and use the corresponding article titles and abstracts only for study context. They are not primary evidence records and do not replace source-level verification. WHO monitoring summaries are generated separately from WHO source reports. An OVE record, by contrast, has a controlled Evidence Type, supporting text, source PMID, structured fields, entity mappings, and an auditable curation state.

06

Controlled names with raw provenance

Virus, host, and country normalization

Authors use abbreviations, historical names, strain codes, informal host descriptions, and locations at different geographic levels. OmniVira preserves the raw wording and links it to a controlled entity only when the mapping is sufficiently supported.

Virus ICTV / VMR

ICTV family, genus, subgenus, and species form the authority backbone. OmniVira adds a curated virus/subtype display layer for named viruses, accepted abbreviations, and influenza HxNy subtypes without replacing the ICTV lineage.

Raw name → confirmed mapping → ICTV hierarchy → curated virus/subtype
Host NCBI Taxonomy

TaxID and scientific lineage remain authoritative. English common names and images are readability aids. Cell systems are kept distinct and may link to their source organism. Host context is classified as Natural host, Experimental system, or Unclear.

Raw name → TaxID / lineage → public display name → host context
Geography Country standard

Specific places and study sites remain available as raw geographic text. Country-level browsing uses canonical English country names and ISO-3 identifiers where a reliable country assignment can be made.

Raw location → country assignment → canonical country / ISO-3

New entities, ambiguous or conflicting mappings, split names, and nonstandardizable phrases remain reviewable. A mapping is not silently promoted merely because an LLM suggests a plausible candidate.

07

Choosing the right entry point

Search, browse, and interpret records

Before interpreting an OVE record

  1. Read the key finding as a structured summary, not as a replacement for the source text.
  2. Open Supporting text and confirm that it directly supports the relevant structured fields.
  3. Check the Evidence Category, Evidence Type, host context, and standardized entities together.
  4. Open the PMID or DOI before reusing a scientific claim.
How abstract highlighting should be read

Supporting-text highlights indicate the abstract sentence or sentence segment most closely aligned with an OVE record. Virus, host, and location underlines are matched primarily from raw extracted names so that later normalization changes do not break provenance. Approximate sentence matching can be used when extraction contains a close paraphrase or merges adjacent source statements.

08

A separate outbreak-monitoring stream

WHO Disease Outbreak News monitoring

WHO monitoring records complement literature-derived OVE records but do not replace them. Each item remains linked to the original WHO Disease Outbreak News report.

WHO source
Title, report date, source URL, and source text obtained from the WHO report.
Public summary
An editable LLM-assisted synopsis intended for rapid orientation.
Key signal
The principal outbreak-related or zoonotic feature identified for monitoring.
Entity links
Standardized virus, host, and country links reviewed through the same entity framework.
09

Continuous but reviewable updating

Daily updates, versions, and quality control

The literature workflow is scheduled daily. A successful run updates the database activity date even when no eligible new publication is found, so “Last updated” describes execution of the update workflow rather than a guarantee of new records.

Retrieve Screen Extract Normalize Validate Queue for review

Daily statistics are organized by PubMed publication or indexing dates where specified, not simply by the time a row was inserted into OmniVira. Screening decisions, exclusion reasons, raw mappings, prompt and schema versions, extraction metadata, and curator actions are retained to support reassessment.

Evidence ontology
v2.1.0
Documentation
2026.07
Documentation reviewed
31 July 2026
Latest recorded data update
12 August 2026
10

Reproducible reuse

Data availability and citation

The public website can be searched without login. Versioned bulk-download endpoints are still being prepared; until they are released, dataset requests can be submitted through the feedback channel.

Primary database article

Manuscript in preparation

The OmniVira database article is currently in preparation. The formal citation and DOI will be provided here when the article is published.

11

Responsible interpretation

Limitations and known sources of uncertainty

  • Abstract-level coverage Evidence extraction may rely on PubMed abstracts when full text is unavailable, so methods and contextual qualifiers can be incomplete.
  • Retrieval bias Coverage reflects the configured PubMed strategy and supplementation rules; relevant literature outside indexed or searchable sources may be missed.
  • Extraction uncertainty LLMs can omit, merge, or misclassify findings. Supporting text and the source publication remain the verification standard.
  • Taxonomic resolution An entity can be standardized at family, genus, species, virus, subtype, or another supported rank. Parent-level mapping does not imply species-level certainty.
  • Association is not causation Detection, seropositivity, receptor compatibility, or phylogenetic similarity alone does not establish reservoir status or a transmission event.
  • Counts are descriptive Evidence and source-article counts describe database coverage, not prevalence, effect size, evidence quality, or public-health risk.
  • Active beta curation Records and mappings can change as curators review new evidence, resolve names, apply ontology revisions, or respond to corrections.
12

Shared vocabulary

Glossary, corrections, and feedback

Evidence record
One structured, traceable finding derived from a source publication and assigned an OVE identifier.
Key finding
A concise statement describing the scientific result represented by one OVE record.
Supporting text
The specific source sentence or close textual support for the structured finding and its core fields.
Raw name
A virus, host, or location expression preserved as it appeared in the source material.
Standardized entity
A controlled virus, host, or geographic record linked to one or more raw mentions.
Spillover
In OmniVira V2.1, a natural or epidemiologically supported non-human animal-to-human transmission event.
Spillback
Human-to-animal transmission, represented under Cross-species Transmission.
Host context
The role of a standardized host mention: Natural host, Experimental system, or Unclear.

To report an error, include the OVE identifier or PMID, the affected virus, host, country, Evidence Type or field, and the source text supporting the proposed correction. This makes the issue reproducible and shortens curator review.