For Researchers

Cohort Exploration

Build cohorts from multi-criteria evidence queries. Combine any evidence types, trust layers, and subject types — including cross-species — in a single query. Every cohort result carries full provenance on each included assertion.

Multi-criteria selection

Cohort criteria are ANDed: a subject must match all criteria to be included. Each criterion specifies an evidence type, optional clinical code, and optional value constraint.

Criterion shape

Field	Type	Description
type	string	Evidence type (e.g. `lab_result`)
code	string?	LOINC, SNOMED CT, ICD-10 code
value	{ lt?, gt?, eq? }?	Numeric value constraint

typescript

const cohort = await client.cohorts.query({
  criteria: [
    // Must have EGFR L858R mutation
    { type: 'genomic_variant', code: 'EGFR_L858R' },
    // AND hemoglobin below 10 g/dL
    { type: 'lab_result', code: 'LOINC:718-7', value: { lt: 10 } },
    // AND NSCLC diagnosis
    { type: 'diagnosis', code: 'SNOMED:254637007' },
  ],
  subjectTypes: ['human', 'canine'],
  trustLayers: ['source_fact', 'normalized_fact'],
});

// Returns:
// {
//   subjects: Subject[]           // Matching subjects with summary stats
//   totalCount: number            // Total matches (before pagination)
//   criteria: CohortCriteria[]    // Echo of applied criteria
// }

Cross-species cohorts

OpenBio's evidence model is species-agnostic. The same hemoglobin assertion type exists for a human patient and a Golden Retriever. Set subjectTypes to include any combination of organism types.

typescript

// Find all subjects — human, canine, and primate — with anemia and NSCLC
const crossSpecies = await client.cohorts.query({
  criteria: [
    { type: 'lab_result', code: 'LOINC:718-7', value: { lt: 10 } },
    { type: 'diagnosis', code: 'SNOMED:254637007' },
  ],
  subjectTypes: ['human', 'canine', 'primate'],
  trustLayers: ['source_fact', 'normalized_fact'],
});

// Each Subject in the result includes:
// { id, name, subjectType, evidenceCount, matchingAssertions }

Export

Export cohort results as JSON or flat CSV for downstream analysis in R, Python, or any statistical tooling.

const cohort = await client.cohorts.query({
  criteria: [
    { type: 'genomic_variant', code: 'EGFR_L858R' },
    { type: 'lab_result', code: 'LOINC:718-7', value: { lt: 10 } },
  ],
  subjectTypes: ['human', 'canine'],
  trustLayers: ['source_fact', 'normalized_fact'],
});

// Export as structured JSON
const json = JSON.stringify(cohort, null, 2);
// → { subjects: Subject[], totalCount: number, criteria: CohortCriteria[] }