GENOME-WIDE ANALYSIS OF 102,084 MIGRAINE CASES IDENTIFIES 123 RISK LOCI AND SUBTYPE-SPECIFIC RISK ALLELES 171 8 statistics for the other study collections except 23andMe are available for bona fide researchers (contact Dale Nyholt, d.nyholt@qut.edu.au) within two weeks from the request. The full GWAS summary statistics for the 23andMe discovery data set will be made available through 23andMe to qualified researchers under an agreement with 23andMe that protects the privacy of the 23andMe participants. Please visit research.23andme.com/collaborate/#publication for more information and to apply to access the data. Code Availability R code for the subtype specificity analysis: https://github.com/mjpirinen/migraine-meta. Results Genome-wide meta-analysis We combined data on 873,341 individuals of European ancestry (102,084 cases and 771,257 controls) from five study collections (Table 1 and Supplementary Table 1) and analyzed 10,843,197 common variants (Methods). Despite different approaches to the ascertainment of migraine cases across the studies, the pairwise genetic correlations were all near 1 (Supplementary Table 2), as determined by LD Score (LDSC) regression,50 showing high genetic and phenotypic similarity across the studies, justifying their meta-analysis. Pairwise LDSC intercepts were all near 0, indicating little or no sample overlap (Supplementary Table 2). The genomic inflation factor (λGC) of the fixed-effect meta-analysis results was 1.33 (Supplementary Figure 1), which is in line with other large meta-analyses69-71and is as expected for a polygenic trait.72 The univariate LDSC51 intercept was 1.05 (s.e. 0.01), which, being close to 1.0, suggests that most of the genome-wide elevation of the association statistics comes from true additive polygenic effects rather than from a confounding bias such as population stratification. The LDSC analysis showed a linear trend between the variant’s LD-score and its association with migraine, as expected from a highly polygenic phenotype such as migraine (Supplementary Figure 2). The SNP-heritability estimate from LDSC was 11.2% (95%CI 10.8-11.6%) on a liability scale when assuming a population prevalence of 16%. We identified 8,117 genome-wide significant (GWS; P < 5 × 10-8) variants represented by 170 LD-independent index variants (r2 < 0.1). We defined the risk loci by including all variants in high LD (r2 > 0.6) with the index variants and merged loci that were closer than 250 kb (Methods).This resulted in 123 independent risk loci (Figure 1, Supplementary Table 3A, and Supplementary Data 1 and 2). Of the 123 loci, 86 are novel whereas 36 overlap with the previously reported 47 autosomal risk loci (Supplementary Table 4) and one with the previously reported X chromosome risk locus. Of the 11 previously reported migraine risk loci that were not GWS in our study, six were GWS in Gormley et al.13 and had P < 3.50 × 10-5 in our data, one had P = 2.37 × 10-3,
RkJQdWJsaXNoZXIy MTk4NDMw