Indirect associations between targets and diseases are computed based on indirect evidence, rather than observed (direct) evidence.

If you search for a target using the web interface of the Open Targets Platform, you will get associations based on direct evidence only. However, if you search for a disease, you will get both direct and indirect associations, since we propagate evidence from lower sections of a disease ontology, this evidence is used as indirect evidence for the associations.

When retrieving our data via the Open Targets REST-API, you will get two types of association counts:

The difference betweentotalanddirectis the number of associations based on indirect evidence.

To get direct associations only, you should include the parameter direct = true when using our REST-API endpoints, such as the association/filter below:

https://api.opentargets.io/v3/platform/public/association/filter?disease=EFO_0000400&direct=true

We propagate the indirect evidence from a child term up to its parent term in the disease ontology for all diseases in our Platform. By computing new target-disease associations when there is no direct (known and observed) evidence for them, we allow:

• finding common targets across groups of related diseases (e.g. ulcerative colitis, Crohn's disease and inflammatory bowel disease)

• making connections between rare and common diseases (e.g. autosomal recessive early-onset inflammatory bowel disease and inflammatory bowel disease)

• grouping evidence for all diseases within a therapeutic area

• identifying unforeseen associations by serendipity

Read our Direct versus indirect evidence: should you care? blog post for more details and email us if you have further questions.

​