Does target-disease evidence from clinical trials include control arm drugs?

Technical Support
1

Some evidence from open target comes from registered clinical trials (Integrated from chembl). Some of the trials are head-to-head competition, i.e, a new drug competing with the best existing drug on the market for a specific disease. I was wondering in the opentarget platform whether both arms will be annotated for a particular trial, or only the new drug will be annotated, as that was the focal drug of the trial?

2

Hi @mengysun!

The clinical trials you see in the Open Targets evidence provide a reference for the link we build between a target and a disease. In this case, the way we approach Clinical trials curation is that we want to collect any proof of the implication for a particular disease when a target is modulated with a drug. So answering to your question, ChEMBL will annotate drugs in all arms of a study, even if the drug present in the control arm is not bringing any novel association.

NCT00320671 is a good example to illustrate this where Aripiprazole is compared against risperidone. If you go to the associations page between schizophrenia and DRD2, which is a common receptor for both molecules with a major role in schizophrenia, you can expect to find the same clinical trial mentioned for Aripiprazole and risperidone.

image
image1505×557 131 KB

In these cases where the amount of references is significant, I’d encourage you to access these records programmatically either by clicking on the API Query button, or by accessing our evidence files:

evd_path = 'gs://open-targets-data-releases/21.11/output/etl/parquet/evidence'

(
    spark.read.parquet(evd_path)
    .filter(F.col('targetId') == 'ENSG00000149295')
    .filter((F.col('drugId') == 'CHEMBL1112') | (F.col('drugId') == 'CHEMBL85'))
    .withColumn('reference', F.explode('urls'))
    .filter(F.col('reference.url').contains('NCT00320671'))
    .select('targetId', 'diseaseId', 'drugId', 'reference')
    .show(truncate=False)
)

+---------------+-----------+----------+------------------------------------------------------------------------+
|targetId       |diseaseId  |drugId    |reference                                                               |
+---------------+-----------+----------+------------------------------------------------------------------------+
|ENSG00000149295|EFO_0000692|CHEMBL85  |{ClinicalTrials, https://clinicaltrials.gov/search?id=%22NCT00320671%22}|
|ENSG00000149295|EFO_0000692|CHEMBL1112|{ClinicalTrials, https://clinicaltrials.gov/search?id=%22NCT00320671%22}|
+---------------+-----------+----------+------------------------------------------------------------------------+

The reason why we want to still capture this known relationship is to amplify the signal that links risperidone, DRD2, and schizophrenia.

I’d be very interested to know if this is the behaviour that you expect to see in the Platform.

Thank you for your question!
Irene

3

Thank you very much! Now I understand how the annotation works, which is helpful for my project. Meanwhile, I am curious if open target plan to integrate more detail informations about these trials. Some commercial database, such as Cortellis or BioMedTracker will follow trials globally from different registries and record the success & failure for individual trials. Are similar steps also within the future plan of open targets?

4

This is definitely something that we are currently considering, so having the input from our users is very valuable. Thank you very much!