The exercise walks through the process of calculating and visualizing Polygenic Risk Scores (PRS) for Coronary Artery Disease (CAD) using publicly available GWAS data and simulated genotype/phenotype data.

This repository contains an exercise from the MiCM Intro to PRS workshop

• The exercise material is adapted from code prepared by Dr. Zoe Schmilovich, June 2024
• The exercise data (European portion of 1000 Genomes) is pre-processed by Dr. Kevin Liang, July 2022

We will be using Google Colab to run our analyses.

• https://colab.research.google.com/drive/1wTGopy_T8LjsdJaevI6swChlBzc3ehm5?usp=sharing.
  

Alternatively, if you would like to run the analyses locally, you will need to have access to the Unix terminal, so be sure you have access to one.

• Mac OS and any linux distribution will have a terminal already
  • Mac OS: search for terminal in your spotlight search
• Windows users:
  • Option 1: Download a unix subsystem (WSL) **recommended
  • Option 2: Download MobaXterm

By completing this exercise, you will:

1. Become familiar with plink2 tool for manipulation of genomic data.
2. Visualize GWAS results with a Manhattan plot and SNP p-value distribution.
3. Perform genotype QC: filtering by MAF, missingness, Hardy–Weinberg equilibrium.
4. LD prune SNPs to ensure independence.
5. Apply precomputed PRS weights to individual-level genotypes.
6. Visualize PRS distribution and test for case–control differences.

Note: This dataset is for educational purposes only. The phenotype labels are simulated and do not represent real disease status.

Link to Google Colab: https://colab.research.google.com/drive/1wTGopy_T8LjsdJaevI6swChlBzc3ehm5?usp=sharing

All material for exercise can be found in the Exercise/ directory

• scripts
  • Step by step instruction for data preprocessing and calculation of PRS
    • This script can be run locally after installing relevant packages (see below) MiCM_PRS_CAD.ipynb
• data
  • CAD GWAS summary statistics (van der Harst et al., 2018) https://doi.org/10.1161/CIRCRESAHA.117.312086
    • raw_cad_gwas_sum_stats is too large and therefore not on GitHub, please see Google Colab documentation for download instructions
  • European genotype subset from 1000 Genomes Project
    • EUR.bim
    • EUR.fam
    • EUR.bed
  • Simulated CAD phenotypes assigned to individuals from the 10000 Genomes Project
  • Intermediate files from quality control
    • CAD_phenotypes.txt
  • Binary files of participants after SNP filtering for genotyping missingness, allele frequency and Hardy-Weinberg equilibrium
    • CAD_maf-0.01_geno-0.1_hwe-1e4.bim
    • CAD_maf-0.01_geno-0.1_hwe-1e4.bed
    • CAD_maf-0.01_geno-0.1_hwe-1e4.fam
  • Binary files of participants after LD-pruning
    • CAD_EUR_QCed.bim
    • CAD_EUR_QCed.bed
    • CAD_EUR_QCed.fam
  • Precomputed PRS weights for CAD
    • CAD_EUR_pst_eff_a1_b0.5_phiauto_allchr.PLINKscore
  • Calculated PRS for individuals
    • prs_for_cad.sscore

Option 1 — Google Colab

1. open Google colab: https://colab.research.google.com/drive/1wTGopy_T8LjsdJaevI6swChlBzc3ehm5?usp=sharing
2. Run all cells in order.

Option 2 — Local Execution

1. Install dependencies:

   pip install gdown pandas numpy matplotlib seaborn

2. Download Exercises/scripts/MiCM_PRS_CAD.ipynb
3. Install locally command line tools:

• [plink2] Documentation: https://www.cog-genomics.org/plink/2.0/
• [gdown] Documentation: https://pypi.org/project/gdown/

Material adapted from Dr. Zoe Schmilovich and Dr. Kevin Liang

Workshop created as part of the McGill Initiative in Computational Medicine