We performed a variance weighted fixed effects meta-analysis using METAL29, testing the effects of 6,037,120 single nucleotide polymorphisms (SNPs) common across all datasets. This revealed a total of 303 genome-wide significant linkage disequilibrium (LD)-independent (r2<0.1) lead variants located in 243 distinct loci. A conditional association analysis using GCTA-COJO30,31 retained 251 independent SNPs in the 243 loci. Manhattan plots are shown in Figure 1A,B, regional plots are provided in Supplementary Figure S1 (X chromosome in Supplementary Figure S2) and details on lead variants are provided in Supplementary Table S2A, while GCTA-COJO independent variants are listed in Supplementary Table S2B. No statistically significant heterogeneity was observed between the datasets (Supplementary Figure S3). 64 of the 243 loci were novel, i.e., not overlapping with the two most recent depression meta-analyses14,25 (Supplementary Figure S4 and Table S2A). The three most significant loci were located near NEGR1, in SORCS3 and in the HIST1 histone cluster, respectively. Among the novel loci, the three strongest associations were in BPTF, LINGO1 and GRIA1 (Table 1). All three genes have been associated with monogenic forms of neurodevelopmental disorders32-34 and
