Human610 quad beadchip

To evaluate the genotyping quality, we compared the concordance rates for the samples genotyped in our study and either (i) samples sequenced by whole-exome sequencing in our previous study1 (link) or (ii) samples genotyped on the Illumina Human610-Quad BeadChip16 (link). For the Exome_Fine Array data, the comparisons were based on 89,720 and 12,320 overlapping variants within 102 and 38 individuals, respectively. The concordance rates were 99.985 and 99.978% for the whole-exome sequencing data and Illumina Human 610-Quad BeadChip data, respectively. Moreover, the concordance rates for the homozygous and heterozygous genotypes were 99.975 and 99.965%, respectively, for the whole-exome sequencing data and 99.874 and 99.954%, respectively, for the Illumina Human 610-Quad BeadChip data. For the Exome_Fine Array data, the comparisons were based on 15,620 and 22,458 overlapping variants within 348 and 159 individuals, respectively. The concordance rates were 99.964 and 99.986% for the whole-exome sequencing data and the Illumina Human610-Quad BeadChip data, respectively. Moreover, the concordance rates for homozygous and heterozygous genotypes were 99.978 and 99.968%, respectively, for the whole-exome sequencing data and 99.865 and 99.976%, respectively, for the Illumina Human 610-Quad BeadChip data. The concordance rate of the 100 blind duplicate samples was 99.988%.

A total of 26,404 FINRISK samples were genotyped using several arrays: the HumanCoreExome BeadChip, the Human610-Quad BeadChip, the Affymetrix6.0, and the Infinium HumanOmniExpress (Illumina Inc, San Diego, CA). The present study, using samples taken in FINRISK in 1997, consisted of 6538 individuals which were genotyped using three genotyping arrays: the HumanCoreExome BeadChip, the Human610-Quad BeadChip, and the Infinium HumanOmniExpress (Illumina Inc, San Diego, CA). Genotype calls were generated together with other available datasets using zCall at the Institute for Molecular Medicine Finland (FIMM). After sample-wise QC (exclude samples with ambiguous gender, missingness [>5%], excess heterozygosity [±4 SD], non-European ancestry) and variant-wise QC (exclude SNPs with high missingness [>2%]), low HWE p-value (<1 x 10^-6), minor allele count (MAC) < 3 (in case Zcall'ed chip data) or MAC < 10 (chip data called using Illumina GenCall) steps, the samples were pre-phased using Eagle2 (version 2.3). Genotype imputation was carried out by using a Finnish population-specific reference panel consisting of 2690 high-coverage WGS and 5092 WES samples with IMPUTE2 (version 2.3.2: Howie et al., 2009 (link)) that allows the usage of two panels at the same time (the ‘merge_ref_panels’ option). Post-imputation QC involved excluding variants imputed with imputation INFO < 0.7.