I am currently utilizing the AUC values from secondary screening in the PRISM database for modeling.
In the paper by Corsello at 2020, AUC values were described as ranging between 0 and 1. I would like to inquire about the interpretation of AUC values greater than 1 in the secondary-screen-dose-response-curve-parameters.csv file.
Can these AUC values greater than 1 represent that the drug is unable to inhibit cell viability or even able to promotes cell growth in the particular cell line? Could these AUC values greater than 1 serve as a negative reference to be provided to regression models for modeling purposes?
Looking forward to your response.
Best regards,
mag
I interpret AUC > 1 as promoting growth, but I would also check the actual dose response curves for any compound-cell line pairs where that occurs to see if it is just noise.
I’m still very curious about how to determine whether the corresponding AUC values greater than 1 have biological significance or are just noise.
For example, in some drug-cell line pairs, the cell viability at all tested concentrations is greater than 2. Although these points may be very scattered, can they represent that the drug does not have an inhibitory effect on that cell line at all these tested concentrations?
If we simply consider data points where cell viability is greater than 1 as noise, then it’s also difficult to believe that data points where cell viability is less than 1 are accurate, isn’t it?
The most crucial point is that I believe these AUC values greater than 1 represent a numerical indication that the drug doesn’t inhibit the cell, thus they can be used in modeling regression models for predicting cell’s drug sensitivity. Do you think this idea might lead to some issues?
Thanks again for your time and efforts on this issue.
Looking forward to your insights~
