Among the brain diseases that we examined, the largest detection gain was obtained with the schizophrenia dataset37, where 18 additional loci met genome-wide significance (excluding those near the MHC region) and were not in linkage disequilibrium (r2<0.2) with the reported susceptibility SNPs37. 7 of these 18 SNPs were found to be associated with eQTLs (Table S8), including rs57709857, which influences LSM1, a gene previously found in a Han Chinese schizophrenia study46. However, the LSM1 locus had not reach genome-wide significance in individuals of European ancestry47. The list of eQTL genes also includes PCNX (associated with rs2189806), a gene encoding a member of the Notch signalling pathway that was reported to harbour a de novo copy number variant linked to Autism Spectrum Disorder48, and CPEB1 (associated with rs1864699), which was recently found to be implicated in experience-dependent neuronal development and circuit formation49 (Figure 5D–E). Thus, several of our new schizophrenia loci have some face validity, but additional replication efforts are required to ensure that these are robust findings. In terms of the percentage increase in detection sensitivity, the largest gain was
