Step 3: Get scores and run metrics script

After running PRS-CS, the output will be one file per chromosome with per-SNP weights. We use the weights.sh script to concatenate all chromosomes together into one file, then use the plink --score function to generate genome-wide polygenic risk scores for each individual in the dataset.

Finally, we use these scores (along with some additional covariates like genotype PCs, age, and sex) as input to the prs_metrics_forWiki.R script. We recommend running this script in Rstudio, and file paths will need to be updated to your directories which holds these files. This script will generate Nagelkerke's R2, liability adjusted Nagelkerke's R2, and AUC (as well as confidence intervals for each) for the cohort. Additional code is provided to generate plots to visualize the difference in scores between cases and controls.