Known_drugs absent from 26.03

Data Access
1

Hi,

Congrats on making yet another release:) In previous releases, there have been parquet downloads for “known_drugs”, e.g.:

If this is intentional (that known_drug is obsolete), could you outline how to obtain the data previously available in “known_drug” with the current parquet files?

Kind regards,

Sigve

2

Hi Sigve!

As you may have noticed in the latest Platform release, the Known Drugs widgets have been retired in favour of the information now consolidated in our Clinical Reports. As a result, this data is organised into separate datasets depending on your starting point. If you’re interested in drug/disease relationships, the Clinical Indication dataset is the right place to look. Similarly, if you want to know whether a target is in the clinic, you’ll want the Clinical Target dataset. As I said, all of these clinical datasets are now derived from the same underlying source, our Clinical Reports; they just present slightly different views based on the research question. I invite you to take a look at our updated documentation to learn more about them!

I also remember you previously raised some issues around drug annotations; hopefully, the new data addresses those points. As always, I’d really appreciate any feedback or suggestions, as we will continue to improve this data.

Have a nice day!

Irene

3

Hi Irene,

Thank you so much for the swift response and the detailed explanations! I was already thinking that there might be data in the clinical reports, I’ll definitely start looking into those to get an overview. And great that you have considered previous feedback on the drug annotations - thanks a lot for all the great stuff you’re doing:)

kind regards,

Sigve

4

Hi Irene,

Been looking at bit more at the latest drug indication annotations after your latest overhaul of the clinical annotations in release 2026.03. Two key observations that makes the user experience somewhat challenged, and that might be relevant to you (although I don’t know exactly how to improve them):

  1. One needs to consider all salt combinations to get the complete list of approved indications (e.g. simply considering TRAMETINIB and not TRAMETINIB DIMETHYL SULFOXIDE will not give a complete picture). I think this is a general pattern for many compounds - and I guess this is something you’re aware of

  2. From what I can see - whether approvals are for single agents or as a combination with another agent is not evident anywhere - this is the case for many targeted cancer agents. It looks like combination approvals are in fact used as the basis for approval status of particular drugs - although maybe not exhaustively (see below). Not sure how these annotations are established in the first place (ChEMBL?), whether it’s literature mining or other sorts of retrieval - but that’s a nuance (combo vs. single) that should be captured moving forward, I think. In relation to this, there are still some indications being missed (I think!.. hard to follow the official approvals too), e.g:

  • Trametinib: NSCLC, glioma, glioblastoma combination approvals show up, but anaplastic thyroid cancer combination approval is missing
  • Encorafenib: melanoma and CRC combination approvals show up, but NSCLC combination approval (with Binimetinib) is missing

I haven’t looked exhaustively at more cases, but would be happy to dig a bit more. I realize this might also be tied to your sources (ChEMBL) rather than your own pipeline, but reckoned it would be valuable to share either way.

best,

Sigve

5

Thanks for your feedback @sigven! We’ll take a look.