CalPred yields calibrated intervals for polygenic risk prediction

Prerequisite Install PLINK2 at https://www.cog-genomics.org/plink/2.0/ Install PRS-CS at https://github.com/getian107/PRScs Install CalPred at https://github.com/KangchengHou/calpred Unzip calpgs.zip to the working directory File structure calpgs.zip Required scripts (see Prerequisite 4). geno_82k_hm3_chr1.* Simulated genotype data used for all analyses, using HapGen2 and 1000 Genome Project simulation_wocontext.ipynb Reproduces simulations without context-specific effects (Figure 1). simulation_wcontext.ipynb Reproduces simulations with context-specific effects (Figure 2). Each simulation notebook performs: Simulation of genetic values Simulation of phenotypes GWAS PGS construction Evaluation of CalPred and PredInterval Input: Simulated genotype data (geno_82k_hm3_chr1.*) or df_woContext.zip or df_wContext.zip (pre-generated intermediate data from PGS construction) i.e., start directly at Evaluation of CalPred and PredInterval. Output: result_woContext.zip and result_woContext.zip These contain pre-generated evaluation results and should match results obtained by rerunning the pipeline. plot_wocontext.ipynb and plot_wcontext.ipynb Reproduce Figures 1 and 2 using outputs from the simulation notebooks or the provided result files. Input result_woContext.zip and result_woContext.zip Replication Software version: python/3.12 R/4.4 plink/2.0-20240804 PRS-CS, commit 5330390 CalPred release v0.1.1 Random seeds are fixed in the document, and replication should be deterministic. In simulation notebooks, comments requesting parallel execution are added; please run the code in parallel. Otherwise, please use the pre-generated intermediate data or pre-generated evaluation results.