Samples were genotyped and imputed as part of the full LIBD cohort, using procedures described previously10,12,13. Briefly, samples were genotyped on four different types of Illumina microarrays over the years (HumanHap650, Human1M, HumanOmni2.5 or HumanOmni5-Quad BeadChips). We merged samples genotyped by the same type of microarray and followed standard preimputation quality control (QC) to remove low-quality (Hardy–Weinberg equilibrium P < 1 × 10−6) and low-frequency (minor allele frequency (MAF) < 0.005) variants. We converted genotype positions from hg19 to hg38 with LiftOver54. Once converted, we imputed genotypes, separately according to genotyping array, on the TOPMed imputation server8,55,56 using the Haplotype Reference Consortium reference panels. We phased genotypes per chromosome using eagle (v.2.4)57. We performed post-imputation QC of each imputed dataset for Black and non-Hispanic WA samples separately.
