Very different magnitude of values for AUC Values

Q&A
1

Hi!
Looking the AUC values of Drug Sensitivity AUC (PRISM Repurposing Secondary) v/s Drug Sensitivity AUC (CTD^2), the magnitudes of the values are extremely different. Drug Sensitivity AUC (PRISM Repurposing Secondary) values are much similar to datasets like GDSC2

Why is that?

2

Could be that the values from Drug Sensitivity AUC (CTD^2) are computed without a fitted logistic regression?

Thanks!!

3

Hi mavergara,

Yes, CTD2 AUC values are the ones originally published in the associated paper and they are not normalized to the covered dose ranges, while for GDSC and PRISM we refit the curves by normalizing the computed AUC’s to the measured dose range (i.e. AUC values are typically between 0 and 1).

You can normalize them by dividing them to their maximum value (16 for most compounds).

4

Hi Mustafa,

Hope you are doing well. Regarding CTD2, I am aware that I would have to normalise the AUC values. I do not seem to understand what you mean by dividing them? Do you mean that we should use min-max normalization technique? I have seen other papers using Z-score technique as well. I really hope you can clarify my confusion. Thank you.