Here is a very similar question to my previous thread: "Scaling conventions for TSC2 predictions" but for ATP7B:
For variants on the ATP7B protein, the validation script prohibits negative predicted values.
I understand that, as explained in the Wiki, "0 = no growth, 1 = wildtype, and >1 denotes improved fitness".
However, experimental distribution provided as response to thread "ATP7B Training Data Request" displays some values below zero.
As you did for TSC2, could you please remove the condition that prohibits negative prediction values for ATP7B?
The question eventually generalizes to other assays like BARD1, LPL and ARSA.
Thank you and all the best,
Matsvei Tsishyn
Created by Matsvei Tsishyn MatsveiTsishyn Apologies for my follow-up question, please ignore it, it was a mistake. Great, thank you very much. Hi @MatsveiTsishyn
Thanks for pointing this out! You’re correct that some experimental distributions (including ATP7B, as well as assays like BARD1, LPL, and ARSA) can contain values below zero. We’ve updated the validation script so that negative prediction values are now accepted, similar to the modification we made previously for TSC2.
This should make your submissions run without the need to clip predictions at zero. Please let us know if you run into any issues with the updated script.
Drop files to upload
Scaling conventions for ATP7B (and others?) predictions page is loading…