Here we provide details of the pre-computed SNP-SNP correlations that we provide for use with GigaPRS. In total, there are ten sets of correlations, but we expect the main four sets will be sufficient for most analyses (the additional sets are useful if analyzing GWAS results from less commonly-used ancestries, or if wishing to avoid using UK Biobank data).
The correlations were constructed using data from the 1000 Genome Project (1000GP) the UK Biobank (UKBB) or the Haplotype Reference Consortium (HRC). They span the five major ancestry groups (African, East Asian, European, Hispanic and South Asian; click here to see a principal component plot). Each set is constructed using approximately 1M autosomal, biallelic SNPs present in HapMap3, and with minor allele frequency (MAF) greater than 0.01 (note that the MAF filtering is ancestry-specific, so the SNP sets vary slightly across sets of correlations).
Click here to see details of the HapMap3 SNPs.
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
Four main sets of correlations:
CARAFR Correlations (from 6578 UKBB individuals; use with African GWAS)
EAS Correlations (from 504 1000GP individuals; use with East Asian GWAS)
INDPAK Correlations (from 6706 UKBB individuals; use with South Asian GWAS)
UK Correlations (from 9880 UKBB individuals; use with European GWAS)
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
Six additional sets of correlations:
AFR Correlations (from 661 1000GP individuals; use with African GWAS)
AMR Correlations (from 347 1000GP individuals; use with Hispanic GWAS)
CEU Correlations (from 5994 HRC individuals; use with European GWAS)
CHI Correlations (from 10028 HRC individuals; use with East Asian GWAS)
FIN Correlations (from 3256 HRC individuals; use with Finnish GWAS)
SAS Correlations (from 489 UKBB individuals; use with South Asian GWAS)
