Association of candidate polygenic scores with Alzheimer’s Disease in UK Biobank validation set.

To select the global tuning parameter, six candidate scores were assessed in a validation set of 119,248 randomly-selected participants of European ancestry from the UK Biobank of whom 279 (0.2%) had been diagnosed with Alzheimer’s disease. Each candidate score was associated with disease in logistic regression models that included age, sex, and principal components of ancestry as covariates and odds ratio (OR) per standard deviation (SD) of polygenic score and area under the receiver operator curve (AUROC) was calculated. The tuning parameter refers to the LDpred ρ parameter used to control the proportion of variants assumed to be causal. Bold indicates polygenic score with maximal AUROC carried forward to the testing datasets. The calibration curves and intercepts were derived by fitting a linear regression model with observed Alzheimer’s prevalence as the outcome variable and predicted prevalence as the independent variable. (XLSX)