Clinical Indication¶

A Clinical Indication represents one drug–disease relationship supported by one or more Clinical Reports. It brings evidence from different records into a compact view while preserving links back to every supporting report.

This is the dataset Open Targets uses to show the diseases for which a drug has clinical evidence. See the Open Targets Indications documentation for its presentation in the Platform.

From reports to an indication¶

Mira starts with individual Clinical Reports. Each report can contain one or more drugs and diseases. Mira forms the drug–disease pairs described by those reports, then groups matching pairs into Clinical Indications.

The example below shows two trial reports supporting the same relationship:

flowchart TD
    CT["<b>Clinical trial</b> <br>NCT02684006 · Phase III</br>"]
    DL["<b>Drug label</b> <br>FDA label · Approval</br>"]
    CI["<b>Curated indication</b><br>ChEMBL · Phase II</br>"]
    DW["<b>Drug warning</b><br>ChEMBL · Withdrawal</br>"]

    CLIN["<b>Clinical indication</b><br>Drug A · Disease B</br>4 contributing reports</br>Sources: trial, label, ChEMBL"]

    APP["<b>Approval</b>"]

    CT --> CLIN
    DL --> CLIN
    CI --> CLIN
    DW --> CLIN
    CLIN --> APP

    subgraph reports["Clinical Reports"]
        direction TB
        CT
        DL
        CI
        DW
    end

    subgraph indication["Clinical indication"]
        CLIN
    end

    subgraph maxstage["Max stage"]
        APP
    end

    style reports fill: transparent, stroke: transparent
    style indication fill: transparent, stroke: transparent
    style maxstage fill: transparent, stroke: transparent

The diagram shows how reports from different origins can support the same drug–disease relationship. The Clinical Indication consolidates them and summarises the maximum stage while retaining the contributing report IDs.

Mapping status¶

Drug and disease identifiers are assigned while Clinical Reports are prepared. A Clinical Indication records how complete that mapping is:

Mapping status	Drug ID	Disease ID
`FULLY_MAPPED`	Known	Known
`DRUG_MAPPED`	Known	Missing
`DISEASE_MAPPED`	Missing	Known
`UNMAPPED`	Missing	Missing

Note

Mira keeps partially mapped and unmapped indications so users can inspect coverage and improve mappings. The Open Targets Platform displays an indication only when both identifiers are known: in Mira, these records have mappingStatus: FULLY_MAPPED.

Important

When an identifier is available, drugName or diseaseName contains that identifier. When mapping is unavailable, it falls back to the source label. These fields therefore act as the grouping keys for the relationship; despite their names, they do not always contain human-readable names.

Maximum clinical stage¶

maxClinicalStage describes the most advanced clinical-development or regulatory stage found across the reports supporting an indication. Mira uses the shared stage ordering described in Clinical Report: Clinical stage.

Note

During current Clinical Indication generation, reports at WITHDRAWAL, PHASE_4, or APPROVAL are represented as APPROVAL before the maximum is selected. The remaining categories follow the clinical-stage order from preapproval through preclinical and unknown.

The maximum stage summarises how far the drug–disease relationship has progressed. Use clinicalReportIds to inspect the underlying evidence and understand which records produced it.

Core fields¶

The core schema is defined in schemas.py. See the Open Targets Clinical Indication schema for the full downloadable dataset contract.

Field	Meaning
`id`	Identifier derived from the drug and disease grouping keys.
`drugName`	Mapped ChEMBL ID when available; otherwise the drug label from the source.
`diseaseName`	Mapped EFO ID when available; otherwise the disease label from the source.
`drugId`	Mapped ChEMBL identifier, or null when the drug is not mapped.
`diseaseId`	Mapped EFO identifier, or null when the disease is not mapped.
`maxClinicalStage`	Most advanced clinical stage across the supporting reports.
`mappingStatus`	Whether the drug, disease, both, or neither have mapped identifiers.
`clinicalReportIds`	Identifiers of all Clinical Reports supporting the relationship.

Example¶

{
    "id": "d91e8ed89d3480f88ff6e4d7389427bd177ce3826dd77d3d174a76919b679ac1",
    "drugName": "CHEMBL1009",
    "diseaseName": "EFO_0004270",
    "maxClinicalStage": "PHASE_3",
    "mappingStatus": "FULLY_MAPPED",
    "clinicalReportIds": ["nct00144209", "nct00625547"],
    "diseaseId": "EFO_0004270",
    "drugId": "CHEMBL1009",
}

This record tells us that:

the drug and disease have both been mapped, so the relationship can be displayed in Open Targets;
drugName and diseaseName use the mapped identifiers;
Phase III is the most advanced stage among the supporting reports;
two Clinical Reports support the relationship and can be retrieved through their IDs.

Trace an indication to its reports¶

clinicalReportIds keeps the aggregated relationship connected to its evidence. For example, to find indications supported by a particular report:

import polars as pl


clinical_indications.filter(
    pl.col("clinicalReportIds").list.contains("nct00144209")
)

The corresponding Clinical Reports contain the provider, source, URL, source-specific metadata, and drug and disease labels used to build the indication.