For this project, we evaluate confounding due to cryptic relatedness in genetic association meta-analysis using simulated data of varying population structure and family relatedness scenarios and real data from SAMAFS and HCHS-SOL. We compare joint analysis, standard fixed-effects meta-analysis (METAL), and our novel method metalcor, which accounts for correlation between studies. Additionally, using simulated kinship matrices, we show that the variance in between-study kinship values empirically scales with substudy sample size and increases with heritability and $F_{st}$.

Folders

• Simulated data results are in sim_scripts folder
• Real data results
  • San Antonio Mexican American Family Studies (SAMAFS) Project 2 (dbGaP accession phs000847.v2.p1) : t2d-samafs_scripts
  • The Hispanic Community Healthy Study/Study of Latinos (HCHS/SOL) (dbGaP accession phs000810.v2.p2) : hchs-sol_scripts
• figures folder : scripts for generating manuscript figures
• data folder : scripts & data for theory figures

Simulated data and real data analysis scripts

• Joint analysis: SAIGE (saige_quant.q and saige_binary.q)
• Sex stratified gwas: (./sex/saige_quant.q and ./sex/saige_binary.q)
• Sex-stratified meta-analysis: (./METAL/meta_sex.q and ./metalcor/)

Simulated traits with real data genotypes

t2d-samafs_scripts/sim_traits/ and hchs-sol_scripts/sim_traits/

• draw_trait.q, draw_trait.R: draw simulated traits with m_causal variants and heritability of 0.8. This script also creates covariate file with PCs for SAIGE
• causal_snp_masking.R: masking causal SNPs with a window of 1000000 (can be adjusted)
• saige_quant.q, saige_quant_male_step1.q, saige_quant_male_step2.q, saige_quant_female_step1.q, saige_quant_female_step2.q : option for LOCO or no LOCO
• combine_chrom_loco.q: for LOCO results, outputs are written per chromosome and need to be combined before evaluation.
• create_metal_file.q, meta_sex.q: generate sex-meta analysis scripts for METAL and run sex-meta analysis
• Final evaluation results/figures: figures/simtrait_LOCO.Rmd