Genotyping and Imputation
UKHLS samples were genotyped with the Illumina Infinium HumanCoreExome BeadChip Kit as previously described (12v1-0).35 This array contains a set of >250,000 highly informative genome-wide tagging single-nucleotide polymorphisms (SNPs) as well as a panel of functional (protein-altering) exonic markers, including a large proportion of low-frequency (MAF 1%–5%) and rare (MAF < 1%) variants. Genotype calling was performed with the gencall algorithm within GenomeStudio (Illumina). After only the samples with matched DNAm data were selected, variants were refiltered prior to imputation. PLINK36 was used for removing samples with >5% missing data. We also excluded SNPs characterized by >5% missing values, a Hardy-Weinberg equilibrium p value < 0.001, and a minor-allele frequency of <5%. For identification of related samples, SNPs underwent LD pruning, and the –genome command in PLINK was used for calculating the proportion of identity-by-descent for all pairs of samples; 58 pairs of related samples (PI_HAT > 0.2) were identified, and randomly excluding one individual from each pair ensured that the samples were independent. These data were then imputed with the 1000 Genomes phase 3 version5 reference panels SHAPEIT and minimac3.37 Best-guess genotypes were called, and variants were filtered to those with a minor-allele frequency >0.01 and an INFO score >0.8. Because variants were named using their locations (“chr:pos”) and variant type (SNP/INDEL), duplicate variants were also excluded. Principal components were calculated from the imputed genotype data via GCTA (a tool for genome-wide complex-trait analysis).38 16 samples were identified as being outliers (defined as more than 2 standard deviations from the mean) in a scatterplot of the first two principal components and were excluded from subsequent genetic analyses. Principal components were then recalculated for inclusion as covariates in QTL analyses. The imputed genetic variants were then filtered so that variants characterized by >5% missing values, a Hardy-Weinberg equilibrium p value <0.001, a minor-allele frequency of <5%, and a minimum of five observations in each genotype group were excluded. These genotype data are available on application through the European Genome-phenome Archive under accession code EGAS00001001232.