type to search

STITCH and, for instance, DrugBank consistency on protein interactions?

Asked by , Edited by Egon Willighagen [ Admin ]

Hi guys,

as I said some time ago, I’m a bioinformatician, but I like hanging around this forum, I find it really interesting. I feel our interests will do nothing but converge in the near future.

Here is my question:

I have a set of 800 FDA-approved drugs. If I go to DrugBank, I find that, on average, these drugs target 3 proteins. However, if I use STITCH 3 with human proteins and a confidence score above 900 (regarded as highest confidence), I find that these drugs interact, on average, with 27 proteins (!).

I’ve found some papers stating that, on average, the polypharmacology of drugs is around 6 targets. Then, why STITCH 3 is reporting so many interactions? A ‘target’ is different than a ‘protein’?

Sorry guys if this question is naive.

NN comments

The title is not a question. Can you correct it?

or Cancel

1 answer


chem-bla-ics [ Admin ]

A ‘target’ is a protein, but a protein is not necessarily a target. The target term comes from drug discovery and a protein which behavior changes when the ligand binds to it, for example, by binding to the active site, changing it’s metabolic or signalling properties. However, it’s not hard to imagine there are many other ways a ligand can interact with a protein. Now, you have to consider what protein-ligand interactions a database is focusing at, and then consider your question again.

or Cancel

Your answer

You need to join Blue Obelisk eXchange to complete this action, click here to do so.