Hello Biostar, I am using the R package PredictABEL which includes functions to assess the performance of risk models. One of the specific function of this package is to produce c-statistics (area under curve of a receiver operating characteristic (ROC) curve) using genetics predictors, in my case SNPs, in a generalized linear model. The software seems to be working with a few SNPs (AUC changing whether I used associated SNPs or not) but when I start using lots of SNPs (~1000) the AUC is always 1 regardless of the SNPs I use. Is it a normal behavior of the model or am I using the package incorrectly.

Thank you François