Golden gate genotypingpanel

We selected 42 candidate SNPs based on associations with either cHOMA-B
[31 SNPs, S1
Table] or arsenic metabolism [11 SNPs, S2 Table] (Balakrishnan et al., 2017 (link)) in the SHFS.
Genotyping was done with DNA from blood collected at baseline using Illumina Infinium
Cardio-Metabo DNA Analysis BeadChip [MetaboChip (Voight et al., 2012 (link)) supplemented with a Golden Gate genotyping
panel (Illumina) containing 670 candidate SNPs for arsenic metabolism and toxicity.
Family-based imputation of genotyped SNPs was done with a PEDSYS-compatible version of
Merlin using human genome build 18 [NCBI36/hg18] and 1000 Genomes as the
reference panel (Abecasis et al.,
2002). Quality control of genetic data excluded participants with call rate
<95%, outliers in identity-by-descent [IBD] clustering, or
outliers in principal components analysis. We excluded SNPs with minor allele frequency
[MAF] <1%, call rate <98%, that were not
autosomal, that were not polymorphic, or that violated Hardy-Weinberg equilibrium
[HWE] P < 10⁻⁵.

Baseline samples were used for single nucleotide polymorphism (SNP)
genotyping using the Illumina Cardio-Metabo DNA Analysis BeadChip (MetaboChip),
a multiplex panel of approximately 200,000 SNPs. The SNPs were selected to
contain all common variants from previous GWAS studies of diabetes, obesity, and
cardio-metabolic diseases and also less common variants not covered on GWAS
chips (MacArthur et al. 2017 ). In
addition, we had available data from a separately genotyped custom panel
(GoldenGate, genotyping panel, Illumina) of 1152 SNPs related to metal toxicity
including scavengers and other genes involved in metal metabolism and
transport.
Detailed information about genotyping quality control and SNP exclusion
criteria can be found elsewhere (Balakrishnan et
al. 2017). Briefly, poorly-performing DNA samples, genotyping
inconsistencies (mendelian errors) were excluded from the analysis. SNPs not
meeting the quality criteria of call rate and minor allele frequency were also
removed for the analyses. Genetic principal components were estimated to address
population stratification by using a subset of unlinked and ancestry informative
markers in unrelated individuals (Price et al.
2006) that were included in subsequent association analyses.