Database Snapshot

What this measures: Counts the total number of each major entity type (AOPs, Key Events, Key Event Relationships, Stressors) and unique authors in the selected version of the AOP-Wiki RDF database. Provides a high-level overview of database size and composition.

Data source: SPARQL queries against the AOP-Wiki RDF endpoint count instances of each entity class (aopo:AdverseOutcomePathway, aopo:KeyEvent, aopo:KeyEventRelationship, nci:C54571 for Stressors) and distinct creator URIs within a single named graph version.

Known limitations: Data reflects the RDF transformation of AOP-Wiki, not the raw database. Author counts are based on unique dc:creator URIs and may not correspond exactly to AOP-Wiki user accounts. Entities without RDF type annotations would not be counted.

SELECT ?graph (COUNT(?aop) AS ?count)
WHERE {
  GRAPH ?graph {
    ?aop a aopo:AdverseOutcomePathway .
  }
}
GROUP BY ?graph
ORDER BY DESC(?graph) LIMIT 1

What this measures: Analyzes the connectivity of AOPs based on shared Key Events. Connected AOPs share at least one Key Event with other AOPs, while isolated AOPs have entirely unique Key Events. This reveals the degree of overlap and interconnection in the AOP knowledge base.

Data source: Two SPARQL queries: one counts total AOPs, the other counts AOPs that share at least one Key Event with another AOP (via aopo:has_key_event). The difference gives the isolated AOP count.

Known limitations: Connectivity is determined solely by shared Key Events declared via aopo:has_key_event. Implicit relationships or shared stressors are not considered. Network density depends on explicitly declared KE-KER relationships in the RDF data.

SELECT ?graph (COUNT(DISTINCT ?aop1) AS ?connected_aops)
WHERE {
  GRAPH ?graph {
    ?aop1 aopo:has_key_event ?ke .
    ?aop2 aopo:has_key_event ?ke .
    FILTER(?aop1 != ?aop2)
  }
}
GROUP BY ?graph
ORDER BY DESC(?graph) LIMIT 1

What this measures: Calculates the average number of Key Events and Key Event Relationships per AOP in the selected database version. This metric indicates the typical structural complexity of AOPs.

Data source: SPARQL queries count the total number of AOP-to-KE and AOP-to-KER relationships (via aopo:has_key_event and aopo:has_key_event_relationship), then divide by the total number of AOPs.

Known limitations: Averages can be skewed by a few very large or very small AOPs. The metric does not distinguish between well-developed and stub AOPs. Relationships must be explicitly declared in RDF.

SELECT ?graph
  (COUNT(?aop) AS ?count)
  (COUNT(?ke) AS ?ke_count)
  (COUNT(?ker) AS ?ker_count)
WHERE {
  GRAPH ?graph {
    ?aop a aopo:AdverseOutcomePathway .
    OPTIONAL { ?aop aopo:has_key_event ?ke . }
    OPTIONAL { ?aop aopo:has_key_event_relationship ?ker . }
  }
}
GROUP BY ?graph
ORDER BY DESC(?graph) LIMIT 1

What this measures: Shows the distribution of Key Event Component annotations (Biological Process, Biological Object, and Action) across all Key Events. Each KE can have up to three component annotations describing the biological context.

Data source: SPARQL queries count Key Events that have annotations for each component type: biological processes (via aopo:has_biological_process), biological objects (via aopo:has_biological_object), and actions (via aopo:has_action).

Known limitations: Only counts presence/absence of each component type, not the quality or specificity of annotations. KEs without any component annotations are not shown. Some KEs may have multiple annotations of the same type.

SELECT ?graph
  (COUNT(DISTINCT ?process) AS ?processes)
  (COUNT(DISTINCT ?object) AS ?objects)
  (COUNT(DISTINCT ?action) AS ?actions)
WHERE {
  GRAPH ?graph {
    ?ke a aopo:KeyEvent .
    OPTIONAL { ?ke aopo:has_biological_process ?process . }
    OPTIONAL { ?ke aopo:has_biological_object ?object . }
    OPTIONAL { ?ke aopo:has_action ?action . }
  }
}
GROUP BY ?graph
ORDER BY DESC(?graph) LIMIT 1

What this measures: Measures how thoroughly each Key Event is annotated by counting how many of the three possible KE Component types (Process, Object, Action) each KE has. Shows the distribution across 0, 1, 2, or 3 annotations per KE.

Data source: SPARQL query counts distinct component types per KE using OPTIONAL patterns for each component type (biological process, biological object, action), then groups KEs by their annotation count.

Known limitations: Treats all component types equally regardless of scientific importance. A KE with 3 low-quality annotations ranks higher than one with 1 highly specific annotation. Only measures breadth, not depth of annotation.

SELECT ?graph ?ke
  (COUNT(DISTINCT ?process) AS ?p)
  (COUNT(DISTINCT ?object) AS ?o)
  (COUNT(DISTINCT ?action) AS ?a)
WHERE {
  GRAPH ?graph {
    ?ke a aopo:KeyEvent .
    OPTIONAL { ?ke aopo:has_biological_process ?process . }
    OPTIONAL { ?ke aopo:has_biological_object ?object . }
    OPTIONAL { ?ke aopo:has_action ?action . }
  }
}
GROUP BY ?graph ?ke
ORDER BY DESC(?graph)

What this measures: Distribution of Key Events across biological levels of organization (molecular, cellular, tissue, organ, organism, population). Reveals which biological scales are most represented in the AOP-Wiki knowledge base.

Data source: SPARQL query counts distinct Key Events grouped by their Level of Biological Organisation annotation (NCI Thesaurus property C25664). KEs without this annotation appear as 'Not Annotated'.

Known limitations: Not all Key Events have biological level annotations. Unannotated KEs appear as 'Not Annotated' and may dominate the chart. The level vocabulary depends on how curators annotated each KE. Coverage varies by version as curation progresses.

SELECT ?level_label (COUNT(DISTINCT ?ke) AS ?count)
WHERE {
  GRAPH <{graph}> {
    ?ke a aopo:KeyEvent .
    OPTIONAL { ?ke <C25664_URI> ?level_obj . }
    BIND(IF(BOUND(?level_obj), STR(?level_obj), 'Not Annotated') AS ?level_label)
  }
}
GROUP BY ?level_label

What this measures: Shows which taxonomic groups (species or organisms) are most represented across AOPs via the Taxonomic Applicability annotation. Identifies the most commonly studied organisms in AOP-Wiki.

Data source: SPARQL query counts distinct AOPs grouped by their Taxonomic Applicability annotation (NCBI Taxonomy property 131567). Only AOPs with explicit taxonomy annotations are included.

Known limitations: Only AOPs with explicit taxonomic applicability annotations are shown. Many AOPs may lack this annotation entirely. Taxonomy labels are extracted from URIs and may appear as identifiers rather than common species names. A single AOP can have multiple taxonomic applicability annotations.

SELECT ?taxon_label (COUNT(DISTINCT ?aop) AS ?aop_count)
WHERE {
  GRAPH <{graph}> {
    ?aop a aopo:AdverseOutcomePathway .
    ?aop <NCBITAXON/131567> ?taxon_obj .
    BIND(STR(?taxon_obj) AS ?taxon_label)
  }
}
GROUP BY ?taxon_label
ORDER BY DESC(?aop_count)
LIMIT 25

What this measures: Histogram showing the distribution of how many AOPs each Key Event belongs to. Most KEs appear in only 1 AOP, but some are shared across many pathways, indicating shared biological mechanisms.

Data source: SPARQL query counts distinct AOPs per Key Event, then bins the results into discrete categories: 1, 2, 3, 4, 5, 6-10, and 11+ AOPs per KE.

Known limitations: Bins KEs into discrete groups for readability. The '6-10' and '11+' bins aggregate multiple values. Only KEs that are linked to at least one AOP via aopo:has_key_event are included. Orphan KEs (not linked to any AOP) are excluded.

SELECT ?ke (COUNT(DISTINCT ?aop) AS ?aop_count)
WHERE {
  GRAPH <{graph}> {
    ?aop a aopo:AdverseOutcomePathway ;
         aopo:has_key_event ?ke .
  }
}
GROUP BY ?ke

What this measures: Counts the number of unique ontology terms used per source ontology (GO, CHEBI, UBERON, NCBITAXON, etc.) in KE annotations. Shows the breadth of ontology vocabulary in use.

Data source: SPARQL query retrieves all distinct ontology-linked IRIs from biological process and biological object annotations. Python post-processing classifies each IRI by ontology namespace prefix and counts unique terms per source.

Known limitations: Shows aggregate counts of unique terms per ontology, not coverage vs reference ontologies (Phase 5 scope). Classification is based on URI prefix patterns. Terms from unrecognized ontologies appear as 'Other'.

SELECT DISTINCT ?term
WHERE {
  GRAPH <{graph}> {
    { ?s aopo:has_biological_process ?term }
    UNION
    { ?s aopo:has_biological_object ?term }
  }
  FILTER(isIRI(?term))
}

What this measures: Shows the distribution of biological process ontology terms used in Key Event annotations. Identifies which ontologies (e.g., GO, MESH, NCIT) are most commonly referenced for describing biological processes in KE component annotations.

Data source: SPARQL queries retrieve all biological process URIs from KE component annotations (via aopo:has_biological_process), then group them by their ontology source namespace.

Known limitations: Only covers processes explicitly annotated in the RDF data. Ontology classification is based on URI prefix patterns. Some processes may be classified differently across ontologies.

SELECT ?graph ?process (COUNT(?process) AS ?count)
WHERE {
  GRAPH ?graph {
    ?ke aopo:has_biological_process ?process .
  }
}
GROUP BY ?graph ?process
ORDER BY DESC(?graph)

What this measures: Shows the distribution of biological object ontology terms used in Key Event annotations. Identifies which ontologies are most commonly referenced for describing biological objects (molecules, genes, proteins, etc.) in KE component annotations.

Data source: SPARQL queries retrieve all biological object URIs from KE component annotations (via aopo:has_biological_object), then group them by their ontology source namespace.

Known limitations: Only covers objects explicitly annotated in the RDF data. Ontology classification is based on URI prefix patterns. Multiple ontology terms may refer to the same biological entity.

SELECT ?graph ?object (COUNT(?object) AS ?count)
WHERE {
  GRAPH ?graph {
    ?ke aopo:has_biological_object ?object .
  }
}
GROUP BY ?graph ?object
ORDER BY DESC(?graph)