Characterizing the Genetic Landscape of Major Depression through Multiple-trait and Cross-ancestry GWAS Meta-Analysis of 1,396,021 Individuals

Background Major depression (MD), a common psychiatric disorder, arises from genetic predisposition and environmental exposure. It is urgent to explore the biological explanation and to enhance the prediction performance, based on the integration of genome-wide association studies (GWAS) of multiple ancestries. Methods We introduced a meta-analysis-based strategy, multiple-trait and cross-ancestry (MTCA), to perform a comprehensive study encompassing 439,605 cases and 1,693,431 controls from European (EUR) and East Asian (EAS). Firstly, using MTAG, we constructed MT-EUR and MT-EAS by ten neuropsychiatric disorders, like Alzheimer’s disease. We used FUMA and MESiuSE to perform single- and cross-ancestry fine mapping, respectively. Then, integrating MT-EUR and MT-EAS, we built the MTCA data by inverse variance weight model in METAL. Using the MTCA data, we used five methods to determine the credible genes which are significant in at least four methods, performed drug reutilization by CMap, and identified significant proteins. Finally, based on MTCA data without UK Biobank (UKB), we constructed 16 MD polygenic risk scores (PRS) using single-trait and cross-ancestry methods in PGSFusion. We performed two kinds of downstream analyses of PGS in EUR and EAS UKB individuals: prediction performance and joint analysis. Results Based on MTCA strategy, we identified 217 risk loci, including 24 previously unreported single nucleotide variants (SNVs). By MESiuSE, we highlighted four causal SNVs with potential cross-ancestry signals. Our analytical approach integrated various methods to pinpoint 45 credible genes and 29 proteins, alongside 17 classes of drugs that hold therapeutic promise. Among the identified loci, we defined rs301806 as a significant association with MD ( P _MTCA = 2.09×10 ^− 9 ) and demonstrated regulatory effects on the RERE across five gene methods, influencing MD risk. After evaluation, DBSLMM-lmm (AUC = 0.65) and PRS-CSx (AUC = 0.62) excelled in in- and cross-ancestry MD risk prediction, respectively. In addition, PRS of MD exhibited significant gender-based interactions (P = 6.70×10 ^− 3 ). Conclusions These advancements not only pave the way for fundamental MD research but also enhance the prospects for tailored diagnostics and therapeutics in clinical settings.