Open Bassa1 opened 2 years ago
This is something that we've also noticed. We did a little preliminary investigation just based on the robokop graph earlier this summer. Here are some note slides: https://docs.google.com/presentation/d/1D-G5PfliiDTteaDyc9hT8jcoJwZMjxlo9YwI_ohoQR4/edit#slide=id.g134558bb5ff_0_269
Bottom line: it's pretty common, spot checks show the occasional mis-labeling, sometimes biology is complicated, and sometimes it may come down to weighing or annotating what is 'normal' and what is 'unusual' or else what is well and less well supported.
Thank for the explanation, in our case, this is not the side effect of the drug. It is an actual problem the drug does not treat psoriasis. The drug is used to induce a mouse model of the disease. Perhaps we need a new predicate to show that the drug cause/ induces a model of disease. Let me know if you need more clarifications on this.
I think that the next question would be what is the provenance of the treats edge for these drugs?
@gglusman I think Bassa was on your team. Can you let me know if this is still a concern? i'm not super clear what query was run so its a bit hard to replicate
This is still indeed a concern. I just re-tested 'what drugs may treat psoriasis' both in test and in demo. In both cases I get Imiquimod offered as a treatment, while the supporting evidence includes papers that talk about psoriasis that is induced/triggered by the drug.
The lithium result is gone.
Hello, Translator Testing Team My team, Multiomics working on the August question of the month for Psoriasis and Psoriasis Arthritis. We use a Translator to find the different drugs related to Psoriasis without a defined predicate. We found that infliximab, imiquimod, and lithium drugs can cause or treat this medical condition. How one drug can have both effects on one medical condition? Any thought on this? I also attached the screenshot of the Translator result here. Let us know if you need any clarification or have suggestions
Basazin