Computational protocol: Common Genetic Variants near the Brittle Cornea Syndrome Locus ZNF469 Influence the Blinding Disease Risk Factor Central Corneal Thickness

Similar protocols

Protocol publication

[…] Three twin cohorts were recruited from Australia and the UK. The AU twin cohort consisted of two sub-samples, 953 individuals from the Brisbane Adolescent Twin Study (BATS) and 761 individuals from the Twin Eye Study in Tasmania (TEST), making up a whole cohort of 1714 participants from 786 families. A full description of the AU twin cohorts is given in Mackey et al . Twins from the UK were a sub-sample from the cohorts collected at St Thomas' Hospital in London. 1759 people from 1119 families were included in this study. Nearly 90% of the UK samples are adult women. Details of the UK twin cohort are given in Healey et al . CCT was measured in the twin cohorts using ultrasound pachymetry and recorded for both eyes. Measurements were performed using a Tomey SP 2000 (Tomey Corp., Nagoya, Japan) or a DGH Technology (model 500; Scarsdale, NY) pachymeter in the Australian and UK twin cohorts respectively. Twin pairs were measured at the same time of day to avoid bias related to diurnal variation. With little evidence for a significant difference between the left and right eyes (ANOVA p-value = 0.575), the mean CCT value of both eyes was used throughout as our measurement.In the AU twin cohorts, DNA samples extracted from each person were hybridized to the Illumina HumanHap 610W Quad arrays, with the samples from BATS genotyped by deCODE Genetics and the ones from TEST genotyped by the Center for Inherited Disease Research (CIDR). We scrutinized the genotypic data (SNPs) and screened them according to a series of quality control criteria, including minor allele frequency (MAF)≥1%, p-value for Hardy-Weinberg equilibrium test≥10−6, SNP call rate>95% or Illumina Beadstudio GenCall score≥0.7. After cleaning, 530,656 SNPs were left for association testing in AU twin cohorts. The UK samples were partly genotyped on the Illumina Hap610W arrays at CIDR, and partly genotyped on Illumina HumHap 300K Duo arrays at Wellcome Trust Sanger Institute. Slightly different quality control criteria compared with the AU twin study were applied: MAF≥1%, p-value for Hardy-Weinberg equilibrium test≥10−4 and SNP call rate>95%, resulting in a complete set of 548,001 SNPs for the association tests.We screened the genotypic data for ancestral outliers using principal component analysis . By comparing AU twin data with 16 global populations sourced from HapMap Phase 3 and Northern European sub-populations from a previous study by McEvoy , 2% of the samples were excluded for being identified as ancestral outliers; thus giving us greater confidence in the homogeneity of the study sample (). UK twin samples were also screened for genetic outliers by comparison with the reference of three main populations from HapMap Phase 2. The Q-Q plot () clearly shows the homogeneity of the UK panel except for one data point. The discrepancy between the observed and expected statistics for this variant suggests a potential association signal.Higher density markers on autosomes were also available from imputation. We imputed data using MACH for the AU samples based on a set of 469,117 SNPs which were common to the six Illumina 610K subsamples at QIMR. The imputation for the UK samples were undertaken with reference to HapMap release 22 CEU using IMPUTE version 2 . Each of the imputed datasets contains up to 2.4 million SNPs.Both AU and UK twin cohorts in our study consist of either twin pairs or their close relatives (parents, siblings) in the family. Samples within the family are genetically related, sharing the chromosomal regions of identity-by-descent (IBD). In those regions, the related samples will provide the similar genetic information. Failing to estimate the IBD states will result in an increased false-positive rate in the association tests. To avoid this problem, we conducted the association test (–fastassoc) in MERLIN . It incorporated genetic relatedness between the samples by estimating the IBD prior to the association tests. The AU samples were controlled for both age and gender effects, whilst the predominantly female UK samples were only controlled for age effects. We standardized the trait distribution of CCT to increase the inter-sample compatibility as well as robustness to extreme observations. […]

Pipeline specifications

Software tools Gencall, IMPUTE, Merlin
Application GWAS
Organisms Homo sapiens
Diseases Glaucoma