PRS-imputation-panels

This reposity contains scripts used in the study "The predictive capacity of polygenic risk scores for disease risk is only moderately influenced by imputation panels tailored to the target population", by Levi, Elkon and Shamir (Bioinformatics 2024, DOI: 10.1093/bioinformatics/btae036).

Below is the description of the main source files

• QC data

  • qc_discovery_data.sh: QC of the discovery data
  • qc_discovery_data.sh : QC of the target data

• cross validation (CV) analysis. Note that method can be one of the following strings: pt2 (P+T target set LD),pt3 (P+T EUR/EAS LD),ls (Lassosum),ld (LDpred):

  • calc_prs_cv_<method>.sh: calculates PRS
  • calc_metrics_<method>: calculates the performance of a specific fold
  • run_folds_<method>.sh: executes nested CV

• Figures and tables generation:

  • Notebooks/aggregate_results.ipynb: summarzies the data
  • Notebooks/generate_figures_and_tables.ipynb: explains how to generate figures and plots

Note that the source files were tailored to the specific configuration of the development enviroment.