How it’s built

This page describes the evidence engine behind Portfolio Intelligence System when Open Targets drives the scope. Assessments are reproducible by design: a GraphQL query against the Open Targets Platform returns structured evidence (target, pathways, disease associations, druggability, safety, DepMap, literature, and more). We turn that into narrative and structured outputs—so you get both the evidence and the interpretation. Completed API runs also emit a opportunity-map.json artifact alongside report.md where the report API is used.

Scope: This is technical depth for Offer 1 (Portfolio Intelligence), not the whole company. Visibility & Narrative and Strategic Capital Options use different methods and artifacts.

Flow

1. One query — We run a comprehensive oncology target assessment query (or a variant for your use case) against the Open Targets Platform GraphQL API.
2. Structured data — The API returns JSON: genes, pathways, associations, known drugs, tractability, prioritisation scores, safety, expression, genetic constraint, literature counts, and more.
3. Narrative report — We interpret that data and write the executive summary, strategic points, and recommendations for your scoped engagement.

Same query, same data—you can re-run as evidence updates or have your own team audit the inputs. For the BRCA2 example, the response was generated from oncology-result.json; the narrative was added for executive use.

Data sources (single engagement can chain all three)

Many diligence scopes use only Open Targets. When the question requires it, we chain additional public, auditable inputs in one engagement—same transparency ethos, broader coverage:

1. Open Targets Platform — GraphQL API evidence (this page).
2. FDA Orange Book — Approved drug listings and related context for comparator and precedent narratives when scoped.
3. Open-source structure prediction — Reference AlphaFold 2 pipeline for computational structural hypotheses, documented in the methods appendix; not a substitute for experimental structure.

For the product positioning and deliverables when all three apply, see Structure-informed diligence.

The query

The BRCA2 report used the following GraphQL query (target ID ENSG00000139618). You can run it in the Open Targets GraphQL Playground or in your own environment. For the Platform’s entities and how they are queried (target, disease, drug, variant, study), see The platform → Entities and query space.

# Comprehensive oncology target assessment
# Replace ensemblId for a different target (e.g. BRCA1, EGFR, TP53)
query OncologyTargetAssessment {
  target(ensemblId: "ENSG00000139618") {
    id
    approvedSymbol
    approvedName
    biotype
    genomicLocation { chromosome start end strand }
    pathways { pathwayId pathway topLevelTerm }
    geneOntology { term { id name } aspect evidence geneProduct source }
    functionDescriptions
    hallmarks {
      cancerHallmarks { label description impact pmid }
      attributes { name description pmid }
    }
    tractability { label modality value }
    targetClass { id label level }
    chemicalProbes { id isHighQuality mechanismOfAction probeMinerScore urls { niceName url } }
    tep { name uri therapeuticArea description }
    knownDrugs(size: 20) {
      count uniqueDrugs uniqueDiseases
      rows { drugId prefName drugType diseaseId label phase status mechanismOfAction targetClass }
    }
    associatedDiseases(page: { index: 0, size: 25 }, orderByScore: "score desc") {
      count
      rows { score datatypeScores { id score } datasourceScores { id score } disease { id name therapeuticAreas { id name } } }
    }
    evidences(efoIds: ["EFO_0000616"], size: 10) { count rows { id score datasourceId datatypeId disease { id } target { id } } }
    isEssential
    depMapEssentiality { tissueId tissueName screens { cellLineName diseaseFromSource geneEffect expression } }
    prioritisation { items { key value } }
    safetyLiabilities { event eventId datasource effects { direction dosing } literature url biosamples { tissueLabel cellLabel } }
    expressions { tissue { id label anatomicalSystems organs } rna { level zscore value unit } protein { level reliability } }
    geneticConstraint { constraintType score oe oeLower oeUpper }
    literatureOcurrences(startYear: 2015) { count rows { pmid publicationDate pmcid } }
  }
}

Example target IDs: BRCA1 ENSG00000012048, EGFR ENSG00000146648, KRAS ENSG00000133703, TP53 ENSG00000141510, PIK3CA ENSG00000121879.

The query file is in this repository at scripts/queries/oncology-target-assessment.graphql. To regenerate the report from the API, run the scripts in scripts/ (see scripts/README.md).

Reproducibility

• Same query, same API — Anyone with access to the Open Targets Platform can re-run this query and get the same evidence base.
• Updates — When Open Targets or upstream sources release new data, re-running the query and regenerating the report keeps the assessment current.
• Audit — The methodology is transparent: data from the Platform; interpretation and narrative by Biotica Bio.

For more on the Platform itself, see The platform.

If you want this workflow applied to your program, request a scoped proposal.