Cindy Boer

Genetics of Osteoarthritis Consortium GWAS Meta-Analyses | 183 4.1 cy was checked against the imputation reference (HRC or 1000G; URLs) to identify pos- sible allele coding errors. p-values values were checked to match the corresponding beta values. Cleaned data was used as input for the meta-analysis. Meta-analysis was performed using inverse variance weighting in METAL[80]. Genomic control was per- formed on all datasets, except those which had already carried out genomic-control ad- justments prior to centralized QC and meta-analysis. Genome-wide significance thresh- old was set at p-value<1.3x10 -08 , corrected for multiple testing. For each phenotype we only considered variants reported in at least two cohorts with the same direction of effect with a minimum MAF>=0.0001 in any contributing cohort. Sex-differentiated meta-analysis The meta-analyses and QC steps described above were repeated for males and females separately in a subset of cohorts ( Supplementary Table 4 ). We then combined the resulting association summary statistics to conduct a sex-differentiated test of associa- tion and a test of heterogeneity in allelic effects, as implemented in GWAMA[9, 79]. This method allows for heterogeneity of allelic effects in magnitude and/or direction be- tween males and females and offers substantial gains in power to detect new SNV asso- ciations. The genome-wide significance threshold was set at p-value<1.3x10 -08 , correct- ed for multiple testing. Heterogeneity in allelic effect sizes was assessed with Cochran’s Q statistic and the significance threshold was set at p-value<0.016, corrected for the three independent new signals identified across the 11 osteoarthritis phenotypes. Significance threshold The testing of =11 osteoarthritis phenotypes in this study needed to be taken into ac- count in the interpretation of genome-wide statistical significance. Applying a Bonferro- ni correction would be inherently conservative as this method assumes independence among the tests considered. Therefore, we first used LD Score regression method[16, 17]with genome-wide meta-analysis summary statistics to estimate the genetic corre- lation matrix between the 11 osteoarthritis traits and then calculated the effective num- ber of independent traits ( M eff ) from the eigenvalues λ i of the correlation matrix[81]: M eff =M−∑ i =1M[I(λ i >1)(λ i −1)] For the M =11 osteoarthritis phenotypes in this study, M eff =4.6565. The threshold cor- rected for the effective number of traits to report genome-wide significance is p-val- ue<1.3x10 -08 .