Previous efforts to study risk prediction for complex disorders have assessed the predictive ability of genetic sum scores based on number of risk alleles that have been associated with a particular disorder. The ability of a test to distinguish between individuals with and without a disease is typically assessed based on the test’s sensitivity, or the proportion of individuals with the condition who have a positive result on the test, and specificity, or the proportion of individuals without the condition who test negative. A frequent measure of clinical validity is the receiver operating characteristic (ROC) curve, which plots the sensitivity vs. 1-specificity for every cut-off of a continuous predictor to distinguish between presence and absence of a disease diagnosis. The area under the ROC curve (AUC) for a continuous predictor corresponds to the probability that an individual with the disease would have a higher predictor score than an individual without the disease, and therefore reflects the proportion of individuals classified correctly as cases or controls. An AUC of 0.5 means that the predictor can accurately classify 50% of individuals, or
