Computational protocol: Novel genetic variants associated with lumbar disc degeneration in northern Europeans: a meta-analysis of 4600 subjects

Similar protocols

Protocol publication

[…] FHS subjects were genotyped using Affymetrix GeneCHip Human Mapping 500 K array set (Affymetrix, Santa Clara, CA, USA) and/or the 100 K array set and/or the 50 K array. Methods and quality controls have been described previously.GARP subjects were genotyped using Illumina Human660W Quad BeadChips (HumanHap550v3, HumanHap610; Illumina, San Diego, CA, USA). Genotyping was performed at the genotyping Rotterdam Genotyping Centre. Positive strand genotypes were called by clustering in Genome studio and imputation was performed using IMPUTE software and hapmap phase II V.21. Strict selection criteria were applied to the measured genotypes using a high information content (r2 of >95%) and a minor allele frequency >0.0025. Association analyses were performed using an inhouse developed software package that allows the analyses of family data using all information available in the cases and controls by extending the Cochran-Armitage trend test.RSI and RS3 subjects in the Rotterdam Study sets were genotyped on the HumanHap550v3 (RS1) or HumanHap610 (RS3) Genotyping BeadCHip (Illumina, San Diego, California, USA). The following sample quality control criteria were applied: sample call rate >97.5%, gender mismatch with typed X-linked markers, evidence for DNA contamination in the samples using the mean of the autosomal heterozygosity >0.33, exclusion of duplicates or first-degree relatives identified using Identity by State probabilities and exclusion of outliers (four SD away from the population mean using multidimensional scaling analysis with four principal components). Filtering criteria for imputation are summarised in supplementary table S1.TUK subjects were genotyped using a combination of Illumina arrays (Human Hap300 and the Human Hap610Q). Genotyping was performed by the Wellcome Trust Sanger Institute using the Infinium assay (Illumina) across three genome-wide single nucleotide polymorphism (SNP) sets, as described previously. Genotyping results had been sent to KCL for collation and analysis using the statistical package, STATA (StataCorp). Strict quality control was applied: 314 075 SNPs were retained for analysis (98.7%); 733 were excluded because their call rates were ≤90% and 725 SNPs had minor allele frequency <0.01. In TUK, significant population substructure was excluded using the STRUCTURE program. [...] All analyses were performed on inverse normal transformed summary LDD score as described above. Each study performed GWA analysis for LDD scores using either MACH2QTL ( (RS1 and RS3) or SNPTEST ( (GARP), which use genotype dosage value as continuous additive predictors of LDD score in a linear regression framework, or ProbABEL using an additive genetic model while accounting for relatedness between the members of a family. Analysis of imputed genotype data accounted for uncertainty in each genotype prediction by using either the dosage information from MACH or the genotype probabilities from IMPUTE. [...] Genotypes for 2.5–3 million autosomal SNPs were imputed separately to increase coverage using HapMap V.2 ( as the reference panel. In GARP and TUK, imputation was performed with IMPUTE V.2 and in the other studies with MACH. The common reference panel led to the reporting of results for the positive strand for all cohorts. In addition, allele pairs were compared between cohorts and no detectable strand-flips were found; the minor allele frequency was also compared between datasets. The distributions of β values of the cohorts were found to be similar and therefore suitable for meta-analysis. All directly genotyped or imputed autosomal SNPs having information from more than one study group (n=2 552 511) were included in the meta-analysis. Association results were combined using inverse variance weighted fixed effects meta-analysis using PLINK V.1.06 ( Two meta-analyses were run: the first was unadjusted; the second was adjusted for age and sex as both are known risk factors for LDD and each risk factor was correlated with LDD in each study group. Heterogeneity of estimated effect was expressed using Q (weighted sum of squares) and I2 (ratio of true heterogeneity to total observed variation). SNPs were excluded from the meta-analysis if the cohort-specific imputation quality, as assessed by r2 (MACH) or Information Score (IMPUTE) metric, was <0.40. On this basis, one marker was excluded from the unadjusted association and one from the adjusted association. […]

Pipeline specifications

Application GWAS
Diseases Hamartoma Syndrome, Multiple, Intervertebral Disc Degeneration
Chemicals Nucleotides