Multivariate genome-wide association study of suicidal behaviors in >1.7 million individuals of diverse population descents

BackgroundWhile previous genome-wide association studies (GWAS) identified multiple risk loci for suicide ideation (SI) and suicide attempt (SA), there is still a limited understanding of the genetic predisposition underlying suicidal behaviors in diverse populations. This study aimed to conduct a large-scale investigation of the suicidality spectrum (SP) to generate new insights into its biology and epidemiology. MethodsLeveraging ancestrally diverse participants (SI Ncase/control=179 881/1 013 900; SA Ncase/control=66 867/1 654 798) from the UK Biobank, All of Us Research Program, Million Veteran Program, FinnGen, and Psychiatric Genomics Consortium, we performed GWAS meta-analyses for SI and SA, and a multivariate GWAS of SP. We applied multiple analytical approaches to identify genomic loci associated with suicide traits and to functionally annotate these significant signals. Phenome-wide genetic correlation and genetically informed causal analyses were further conducted to provide convergent evidence on the relationships of suicidal behaviors with health outcomes, brain imaging-derived phenotypes, and metabolomic traits. We identified 90 independent lead single-nucleotide polymorphisms (SNPs) for suicidal behaviors, of which 49 were novel. SNP-based heritability was higher for SA (SNP-h2=0.115{+/-}0.005) than for SI (SNP-h2=0.040{+/-}0.002) and SP (SNP-h2=0.050{+/-}0.002), and their genetic correlations ranged from 0.639 to 0.960. Functional annotation analyses highlighted associated genes (e.g., UGGT2, GMPPB, BRWD1), enriched gene sets (e.g., cellular response to stress), and potential therapeutic drug candidates (e.g., cariprazine, paliperidone, droperidol), with some signals shared across suicide traits but more specific to SI or SA. Suicide behaviors were genetically associated with a broad spectrum of complex traits, primarily in the domains of mental health, physical health, behaviors, and socioeconomic factors, some of which revealed causal relationships. InterpretationThis study provides convergent genetic evidence for both shared and phenotype-specific components of suicidal behaviors and delineates their associated factors spanning from proximal clinical and behavioral traits to more distal social determinants. These findings refine our understanding of the etiology of suicidal behaviors and may inform targeted strategies for suicide prevention in both clinical and public health settings. Research in contextO_ST_ABSEvidence before this studyC_ST_ABSSuicidal behaviors remain a serious public health concern and contribute to substantial mortality globally. However, their genetic predisposition and risk profiles associated with them are not yet fully established although multiple approaches, including genome-wide association studies (GWAS), have been applied. We searched PubMed, medRxiv, and bioRxiv for publications and preprints in English from Jan 1, 2000, to Nov 1, 2025, using the search terms "suicid*" and "GWAS". The largest previously published GWASs identified four genome-wide loci for suicide ideation and 12 for suicide attempt, and a recent preprint reported 77 loci for suicidal behaviors. These associations were mainly derived from individuals of European ancestry, with only one significant locus for suicide ideation reported in East Asian ancestry. Previous studies also attempted to prioritize putative risk genes, but most did not leverage multi-omic approaches, such as transcriptome-wide and proteome-wide association studies. Moreover, earlier work investigated the relationships between suicidal behaviors and psychiatric and behavioral traits mainly through genetic correlation analysis, without conducting phenome-wide genetically informed causal analysis. Added value of this studyOur study identified 90 genome-wide significant associations for suicidality. These associations were distributed across European, African, Admixed American, and Asian ancestries for both suicide ideation and suicide attempt. SNP-based heritability for suicidal behaviors ranged from 4% to 12% and remained significant after conditioning on major psychiatric disorders. We prioritized 1052 genes associated with suicidal behaviors and identified 33 loci with convergent evidence across five or more gene-discovery analyses. Sixteen genes were shared across suicide ideation, suicide attempt, and suicidality spectrum, while most other loci appeared to be phenotype-specific. Drug-repurposing analysis suggested five potential therapeutic drug candidates: cariprazine, droperidol, molindone, paliperidone, and chlorprothixene. Using phenome-wide genetically informed analyses, we identified loneliness, medical abortion, and age at first sexual intercourse as putative causal risk factors for suicidal behaviors. Suicidal behaviors were also associated with adverse consequences, including hospital admissions and multiple mental health and physical conditions. Implications of all the available evidenceThe predisposition to suicidal behaviors is due to genetic mechanisms acting through molecular changes across multiple omic domains in brain and peripheral systems. Although psychiatric disorders may mediate the genetic liability of suicidal behaviors, a proportion of the risk appears to be independent of psychiatric diagnosis. Suicidal behaviors are influenced by diverse genetic and causal risk factors, and can also lead to broad adverse health consequences.