variani/paper-neff

Scripts for the Neff paper

Scripts for the manuscript, Ziyatdinov, et al. "Estimating the effective sample size in association studies of quantitative traits." bioRxiv (2019), https://www.biorxiv.org/content/10.1101/2019.12.15.877217v2.full.

Analysis steps

Step 1. UK Biobank data preparation

• 336K unrelated individuals of British ancestry
• >600K genotyped SNPs, autosomes, MAF >0.1%, QC
• 6 traits: bmi, weight, waist, hip, height, whr
  • impute missing trait values by mean (<1% of missingness)
  • project out covariates: age/sex + PC1-20
  • apply rank-based inverse normal transformation

Scripts to process raw UK Biobank data are not shared.

Step 2. GWAS by Linear Regression (GWAS-LR)

• script: scripts/08-gwas-lm-top.R
• snakemake command: snakemake -s sm.py lm

Step 3: Clumping with p-values from GWAS-LR

• plink command: plink --bfile <bed> --out <out> --clump <assoc> --clump-p1 0.1 --clump-p2 0.1 --clump-r2 0.01
• results: out/clump/

Step 4: Heritability estimation by low-rank LMM

• script: scripts/04-estimate-h2.R
• snakemake command: snakemake -s sm.py h2
• results: out/h2/

Step 5: GWAS by low-rank LMM (GWAS-LMM) using LOCO

• script: scripts/07-gwas-lmm-loco-top.R
• snakemake command: snakemake -s sm.py loco