Why do the AUC from GDSC (0-1) and CTRP (0-30) have different scales?

Q&A
1

Hi.
I have downloaded the GDSC (sanger-gdsc-dose-response.csv) and CTRP (CTRPv2.0_2015_ctd2_ExpandedDataset). I want to use the AUC as the measure of drug response but I found the AUCs have different scales, with CTRP ranging from 0 to 30 and GDSC ranging from 0-1. A possible reason may be one is fraction and the other is absolute value? Is there any way to standardize the AUC from CTRP to enable it to be between 0-1?
Thanks.

2

Hi cjlin,

Yes, your guess for the difference between the scales is correct. We kept the AUC values in the CTRP screen as in the original manuscript since the compounds in that dataset have different dose ranges.

For normalization, you can divide the CTRP AUC values with the corresponding maximum dose for each compound.

Warmly,
Mustafa

3

Hi, could you provide additional information on this. After a bit of investigation of the CTRP data, it does not appear to be true that the AUC values on DepMap are just standard AUC values multiplied by the maximum concentration. Looking at the AUC data on DepMap and taking the maximum AUC for each drug, the values range from 13.574 to 29.350. However, the range of concentrations is from 0.13 - 600 uM. If the AUC data on DepMap was just 0-1 values that were multiplied by the maximum concentration, then the maximum AUCs for each drug should be 1*maximum_concentration. Therefore, the maximum AUC values across the drug should range from 0.13 to 600, not ~13 - 29.

You can even take some examples, CIL56 is listed as having a top testing concentration of 2.5, but the AUC values range from 6.92 to 17.19. Meaning the “original” AUCs would all be greater that 1.