ABSTRACT
Large-scale genome-wide association studies of schizophrenia have uncovered hundreds of associated loci but with extremely limited representation of African diaspora populations. We surveyed electronic health records of 200,000 individuals of African ancestry in the Million Veteran and All of Us Research Programs, and, coupled with genotype-level data from four case-control studies, realized a combined sample size of 13,012 affected and 54,266 unaffected persons. Three genome-wide significant signals — near PLXNA4, PMAIP1, and TRPA1 — are the first to be independently identified in populations of predominantly African ancestry. Joint analyses of African, European, and East Asian ancestries across 86,981 cases and 303,771 controls, yielded 376 distinct autosomal loci, which were refined to 708 putatively causal variants via multi-ancestry fine-mapping. Utilizing single-cell functional genomic data from human brain tissue and two complementary approaches, transcriptome-wide association studies and enhancer-promoter contact mapping, we identified a consensus set of 94 genes across ancestries and pinpointed the specific cell types in which they act. We identified reproducible associations of schizophrenia polygenic risk scores with schizophrenia diagnoses and a range of other mental and physical health problems. Our study addresses a longstanding gap in the generalizability of research findings for schizophrenia across ancestral populations, underlining shared biological underpinnings of schizophrenia across global populations in the presence of broadly divergent risk allele frequencies.
Introduction
Schizophrenia and related psychoses occur in all human populations, but are diagnosed most frequently among those racialized as Black and those of primarily African ancestries1. In the United States, such patients experience worse treatment outcomes (e.g. hospitalization, functioning) as a consequence of racism and other social determinants of health, compounding pervasive inequities which impact morbidity, mortality, and quality of life2,3. Decades of research point resoundingly to a heritable basis of risk, which is shared across diagnoses and populations4,5 and does not account for differences in prevalence and presentation6,7. Furthermore, because the largest genome-wide association studies (GWAS) have overwhelming recruitment bias towards European populations8, the continued underrepresentation of African ancestries in genomics research threatens to limit the benefits of novel biological insights and advances in precision medicine, further entrenching these disparities.
A primary obstacle in genomics research on schizophrenia is its extreme polygenic architecture. Hundreds of loci have been identified to date, which exert very small effects in concert with thousands more that do not attain stringent genome-wide significance. Importantly, prior research indicates that most of the currently attributable risk (i.e. in European populations) resides in variation predating human migrations out of Africa, and is shared across worldwide populations. However, diverging allele frequencies, effect sizes, and patterns of linkage disequilibrium (LD) render polygenic risk scores (PRS) that decline continuously in performance as genetic distance between training and target samples increases9–11. These limitations are especially marked in populations of African ancestry.
Building on our recent work characterizing and validating psychiatric diagnoses from electronic health records (EHRs)12,13, we newly integrate genome-wide single nucleotide polymorphism (SNP) genotyping data for over 124,000 African ancestry individuals (AFR) from the Million Veteran Program (MVP)14 and Cooperative Studies Program (CSP) #57215. We comprehensively assess the replicability of major schizophrenia findings and, for the first time, recapitulate characterizations of the genetic architecture that to date have been limited to populations of European (EUR) and East Asian (EAS) ancestry. Combining results with civilian data from the All of Us (AOU) Research Program16, and four published case-control studies17–20 culminated in a multi-stage GWAS of 13,012 affected persons and 54,266 controls of admixed African ancestries (Supplementary Note). Subsequent meta-analysis with European and East Asian data yielded a total of 86,981 cases and 303,771 controls. An overview of the AFR-specific and cross-ancestry meta-analysis strategies is given in Figure 1A. To increase our understanding of the biological mechanisms underlying schizophrenia across ancestries, we integrated outcomes from the cross-ancestry fine-mapping analysis with single-cell functional genomic data from the human dorsolateral prefrontal cortex.
Results
Schizophrenia genetic associations in AFR populations
We initially compared 6,408 patients with schizophrenia and 20,065 controls enrolled in CSP #572 and MVP (Stage 1), and then extended our analysis to an additional 6,604 cases and 34,201 controls (Stage 2) from AOU16, Consortium on the Genetics of Schizophrenia (COGS)18, Genomic Psychiatry Cohort (GPC)20,21, Molecular Genetics of Schizophrenia (MGS)19, and Project among African-Americans to explore risks for schizophrenia (PAARTNERS) studies17,22 (Extended Data Table 1; Supplementary Data 1).
Our GWAS in CSP #572 and MVP (Stage 1) uncovered six statistically independent genome-wide significant (P<5×10-8) associations with SNPs in ZBTB20 on 3q13.31, BTLA on 3q13.2, GRIN2A on 16p13.2, near WSCD1 and AIPL1 on 17p13.2, downstream of the Homeobox B (HOXB) gene cluster on 17q21.32, and upstream of PMAIP1 on 18q21.32 (Figure 1B; Extended Data Table 2). Of these, only PMAIP1 was directly replicated in the expanded AFR meta-analysis (13,012 cases and 54,266 controls), while two additional regions — upstream of PLXNA4 on 7q32.3 and TRPA1 on 8q21.11 — attained genome-wide significance with this analysis (Figure 1B; Figure 1C).
Notably, none of the specific genome-wide significant associations observed in AFR have been reported in published studies of EUR or EAS populations. However, suggestive AFR findings (P<10-5) included previously replicated associations with SNPs in FAM120A (9q22.31) and DCC (18q21.2) (Supplementary Table 1).
Cross-ancestry GWAS of schizophrenia
We obtained EUR summary statistics from the Psychiatric Genomics Consortium (PGC3; 53,386 cases and 77,258 controls) and combined these with new MVP-EUR results (6,579 cases and 155,490 controls) to yield an expanded GWAS of 59,965 cases and 232,748 controls (Figure 2A - left panel; Extended Data Table 1). We then combined our AFR results (13,012 cases and 54,266 controls) with these expanded EUR results and available EAS summary statistics23 (14,004 cases, 16,757 controls) in cross-ancestry meta-analyses across 18,982,014 individual variants in 86,981 cases and 303,771 controls (Figure 2A - right panel; Extended Data Table 1; Supplementary Data 1).
We resolved 19,529 genome-wide significant associations (excluding 5,021 mapping to the extended MHC region) to 307 statistically independent signals using conditional and joint (COJO) multi-SNP analysis (Methods), which accounts for linkage disequilibrium (LD) among nearby variants (Supplementary Table 2). By comparison, meta-analyses of EUR and EAS, EUR and AFR, and AFR and EAS yielded 298, 257, and 13 independent signals, respectively (Supplementary Tables 3-5). Taken together with within-ancestry results for AFR (3) and EUR (248) (Supplementary Table 6), this corresponded to 708 unique variants spanning 397 physically distinct chromosomal loci (Figure 2B). As expected, there was starkly lower power to detect genome-wide significant associations in AFR compared to EUR populations and only marginally greater power in cross-population analyses of three ancestries (Figure 2C).
We intersected our findings with 270 loci in the most recent PGC study24 using LD-”clumped” intervals, highlighting 151 independent variants in 111 novel loci that were more than 1Mb from a previously reported locus (Figure 2A and Figure 2B; Supplementary Table 7). Multi-ancestry GWAS findings revealed additional novel loci implicating genes, including CACNA1B, KCNN4-AS1, NRP1, PTPRT, SEMA3A, and SLC12A5, which play important functions in neuronal signaling.
We compared directions of allelic effects between ancestries, observing that, for progressively more stringent P-value thresholds (PT) applied to the PGC-EUR findings, the proportion of same-direction effects in EUR replication and EAS results converged to 1 (Figure 2D). As applied to AFR results, in which the same PGC-EUR signals corresponded to a larger effective number of LD-independent SNPs, we observed fractions of same-direction tests exceeded 60% at (Psign<10-3 at PT<10−10); however, this fraction exceeded the null expectation of 50% concordance at all tested thresholds. A direct comparison of LD-independent genome-wide significant SNPs revealed a modest but significant correlation between AFR and EUR effects (ρ=0.209; P=7.55×10-7), despite fewer directionally concordant results than seen with an independent EUR sample (ρ=0.812; P=8.25×10−53) (Figure 2E; Supplementary Table 8).
Enhanced fine-mapping of schizophrenia-associated loci
We evaluated the impact of expanded diversity on fine-mapping resolution at new and previously associated loci. From the COJO SNPs in each meta-analysis, we defined 773 unique intervals separated by at least 250 kilobases (kb), represented by the most significant index SNP in a given window (Supplementary Table 9; Supplementary Data 2-4). For each window, we estimated a 95% credible set of SNPs, corresponding to a genomic interval containing one or more causal variants driving a given association signal. For the EUR-only and the meta-analyzed GWAS, we used SuSiE-R with a “blended” LD matrix (Figure 3A; Methods). The number of causal SNPs we identified (posterior inclusion probability, PIP>0.5 and P<5×10-8) was 51 for EUR only, 245 for AFR+EUR, 83 for EUR+EAS, and 366 for AFR+EUR+EAS (Supplementary Table 10). Additionally, fine-mapping between the unique pairs of distinct ancestries (AFR+EUR, EUR+EAS, AFR+EAS) was performed using MESuSiE (Figure 3A; Supplementary Tables 11-18; Methods). Lastly, for the three distinct ancestries (AFR+EUR+EAS), fine-mapping was conducted with SuSiEx (Figure 3A; Methods), which is less conservative than MESuSiE as it considers all comparisons among ancestries. The number of SNPs with PIP > 0.5 was 248 for AFR+EUR, 41 for EUR+EAS, 68 for AFR+EAS, and 261 for AFR+EUR+EAS (Supplementary Tables 19-20).
We observed higher posterior probabilities following meta-analysis of EUR with EAS and AFR results (Figure 3B) and shorter genomic intervals (kb spanned by credible SNPs) (Figure 3C). In the EUR only analysis, out of 773 unique intervals separated by at least 250 kb, 606 indexes had at least one credible set defined, with 72 SNPs exhibiting a PIP greater than 0.5 and 51 SNPs reaching genome-wide significance (P<5×10-8). In the AFR+EUR analysis, 671 unique indexes were identified, including 1,056 SNPs with a PIP greater than 0.5 and 245 SNPs that were genome-wide significant. For the EUR+EAS analysis, we found 678 unique indexes with at least one credible set, among which 208 SNPs had a PIP greater than 0.5 and 83 SNPs were genome-wide significant. Lastly, the AFR+EUR+EAS analysis revealed 697 unique indexes, with 1,353 SNPs having a PIP greater than 0.5 and 366 SNPs reaching genome-wide significance.
Leveraging local ancestry in admixed AFR populations
We investigated the impact of admixture in AFR individuals through local ancestry deconvolution, followed by joint modeling of AFR and EUR haplotype tracts using an adaptation of the tractor approach25 (Methods). We observed extended genomic regions that harbored significant (P<10-6) associations between the number of AFR haplotypes carried by an individual and reduced disease risk (Extended Data Figure 1A). Comparing standardized effect sizes estimated from AFR tracts and EUR tracts to PGC-EUR results (PT=5×10-8), we observed a stronger correlation with EUR tracts (ρ=0.435; P=1.88×10−11) than AFR tracts (ρ=0.263; P=4.29×10−10) (Extended Data Figure 1B). Similarly, the proportion of same-direction tests was greater for EUR (0.68; Psign=8.67×10-8) than AFR (0.58; Psign=2.35×10-3) tracts (Extended Data Figure 1C; Supplementary Table 8).
Examination of tract-level results revealed distinct AFR-specific evidence of association near PLXNA4 (PAFR=6.75×10-5; PEUR=0.999), PMAIP1 (PAFR=7.22×10-5; PEUR=0.985), and TRPA1 (PAFR=5.56×10-4; PEUR=0.207) (Extended Data Figure 2A-C). We observed even stronger tract-level associations with SNPs in CLEC16A on 16p13.13 (PAFR=3.94×10-7; PEUR=0.6414), DOCK4 on 7q31.1 (PEUR=1.15 ×10-6; PAFR=0.649), and GRIN2B on 12p13.1 (PAFR=8.93×10-8; PEUR=0.696). At GRIN2A, we observed overlapping but distinct associations based on AFR (PAFR=6.10×10-6; PEUR=2.35×10-3) and EUR tracts (PEUR=1.42×10-6; PEUR=0.884) (Extended Data Figure 2D-F).
Taken together, these results suggest the intriguing possibility of population-specific haplotypes or complex patterns of LD that hinder detection of causal schizophrenia-related variation in the presence of local admixture. We explored the potential for enhanced fine-mapping resolution (i.e. beyond standard trans-ancestry meta-analysis) by re-estimating credible sets based on meta-analysis of our overall EUR findings with tract-level AFR statistics. In particular, at the CACNA1I locus, this approach narrowed the corresponding credible set from 64.7kb to 39kb spanning exons 2 and 3 and introns 1-3 (Figure 3D). Within this region, 10 SNPs had a PIP greater than 0.9, compared to a maximum PIP of 0.244 based on EUR results alone.
Comparative estimates of SNP-based heritability
We sought to better characterize the population-specific genetic architectures of schizophrenia in AFR and EUR populations, by leveraging available individual-level genotype data for the combined CSP #572 and MVP cohorts. Using common SNPs with MAF ≥ 1%, we obtained near-identical estimates of SNP-h2 for AFR (0.183±0.013) and EUR (0.183±0.017) participants (Extended Figure 3A; Supplementary Tables 21-22); including variants with MAF greater than 0.1% yielded comparable increases in SNP-h2 for AFR (0.268±0.026) and EUR (0.270±0.019). Adjusting these estimates by the number of SNPs in each population illustrates that, in EUR populations, an equivalent fraction of variability in individual risk is explained by fewer SNPs (Extended Data Figure 3A). In contrast, calculating SNP-h2 from ∼1.2M HapMap3 variants yielded markedly attenuated estimates for AFR (0.138±0.014) but not EUR (0.196±0.012), highlighting the limitation of using “convenience sets” of SNPs optimized for EUR populations 26–28.
Given the extensive genetic diversity present in AFR populations, we partitioned SNPs which were common (MAF>0.1%) in AFR individuals but rare (MAF<0.1%) in EUR, and variants which were common in both. Among AFR individuals, shared common genome-wide variants explained around a quarter of the total attributable variance (0.074±0.015), while less common (MAF<1%) variation explained a 1.5-fold larger fraction of individual variability in liability (0.114±0.027) (Extended Figure 3B; Supplementary Tables 23).
Next, we partitioned SNPs in EUR participants based on the absolute allele frequency difference (AFD) 29 with AFR populations. We found that a negligible proportion of variance was explained by population-”specific” variation (0.014±0.029), with most of the attributable variance explained by SNPs in the third (0.075±0.016) and fourth (0.071±0.013) AFD quartiles (Extended Figures 3C; Supplementary Table 24). Taken together with the observations above, this suggests that variation exhibiting greater divergence between populations harbors a larger fraction of disease-related variance.
Cell-type convergence of EUR and AFR heritability for schizophrenia
Previous studies support the enrichment of EUR schizophrenia risk loci for gene expression markers related to excitatory and inhibitory neurons24,30,31. We examined whether the genetic architecture of schizophrenia in AFR converges to similar cell type enrichment. For this purpose, we used the extensive PsychAD single-cell atlas of 6.3 million nuclei from the human prefrontal cortex32 and applied the scDRS method33 that considers transcriptional heterogeneity between individual cells to calculate polygenic risk score for each cell population (Methods). Consistent with multiple lines of prior evidence 24,30,31, we observed a significant enrichment in all subtypes of excitatory and inhibitory neurons in EUR (Figure 4B). While the EUR and AFR enrichments are well correlated (Pearson’s r = 0.702; Figure 4C), not all significantly enriched neuronal subtypes were replicated in AFR which can be attributed to lower GWAS power in AFR. Out of the significant findings, oligodendrocyte precursor cells (OPCs) were more highly enriched in AFR than EUR GWAS (Figure 4B).
Cell-type transcriptome convergence of EUR and AFR heritability for schizophrenia
We leveraged single-nucleus gene expression data from the prefrontal cortex of 920 EUR and 321 AFR donors that are included in the PsychAD cohort32 to perform an ancestry-stratified transcriptome-wide association study (TWAS). Across 31 cell-type-specific models, EUR models could confidently impute 18,219 unique genes across 112,474 gene-cell-type combinations, while AFR models could confidently impute 16,956 unique genes across 74,002 gene-cell-type combinations (Methods). Integrative analysis of the 31 cell-type-specific EUR models with the expanded EUR GWAS identified 2,057 unique genes (Bonferroni-adjusted P-value < 0.05). Similar analysis using the AFR models with the AFR GWAS identified only 3 unique genes at the same statistical threshold, likely due to the lower AFR GWAS power. To assess concordance between EUR and AFR schizophrenia TWAS, we quantified the gene correlation as a function of top ranked genes (as defined by meta-analysis P-value). Across all well-represented cell-types, we find that correlation among EUR and AFR TWAS is ∼0.7-0.9 for the top ranked genes (Extended Data Figure 4), highlighting the biological convergence capturing genetically driven schizophrenia-associated gene expression shared between EUR and AFR.
Identification of putative causal genes through enhancer-promoter interaction model
Expression quantitative trait loci (eQTL) mapping and colocalization with GWAS variants is a common method for nominating credible causal genes34; however, eQTLs explain only a small fraction of GWAS signals due to their clustering near transcription start sites and association with genes that have simpler regulatory landscapes35. To address this limitation, cell-specific enhancer-promoter (E-P) models36 were trained using single nucleus (sn) multiome (snATAC-seq + snRNA-seq) and fluorescence activated nuclei sorted (FANS) Hi-C data from early postnatal to late adulthood human brains37 (Methods), capturing long-range chromatin interactions and providing an accurate map of regulatory effects on gene expression38. Using these E-P interaction maps, we connected COJO SNPs with putative regulatory sequences and credible causal genes (Extended Data Figure 5A and Methods) to detect 1,538 E-P interactions across all cell types. 767 unique genes that were putatively associated with 40% of the conditionally independent variants (Extended Data Figure 5B; Supplementary Table 25) primarily affect E-P interactions within one cell type (Extended Data Figure 5C).
Despite the differences in methodologies and the sources of omics data, we found a significant overlap of nominated causal genes predicted by the E-P and TWAS approaches for all major cell types except endothelial cells, for which TWAS reported a markedly lower number of causal genes due to the low frequency of these cells in the training dataset (Bonferroni P-value between 7.15×10−36-3.30×10-6). A total of 94 prioritized protein-coding identifiers were identified across all cell types (Figure 4D; Supplementary Table 25). In line with prior research, our analysis revealed that these genes are significantly associated with abnormalities in both presynaptic and postsynaptic regions (Figure 4E) as they participate in 11 ontological sets in the synaptic organization, covering a variety of specific pathways and functions (Supplementary Table 26). For instance, four and three of the prioritized genes encode components of the postsynaptic density membrane (adjusted P-value=0.00175) and the presynaptic active zone membrane (adjusted P-value=0.00175), respectively, highlighting the disruption of several vital processes essential for synaptic transmission.
Among the 94 shared genes, we discovered a connection between schizophrenia COJO GWAS variants rs57022825 and rs17486822 located within a glutamatergic-specific enhancer inside the intronic region of the NALCN gene (Sodium Leak Channel Non-Selective Protein; Figure 4F). NALCN is a major player in determining the influence of extracellular Na+ on a neuron’s basal excitability and its modulation by hormones and neurotransmitters39. Although NALCN variants have been reported to be in linkage disequilibrium with schizophrenia and bipolar disorder40,41, this is the first evidence revealing the cell-specificity of that regulation, providing strong support as it comes from the consensus of two independent methods. Another example is a novel link between an LD buddy of schizophrenia COJO SNP rs6065926, situated within a microglial enhancer, and the cluster of differentiation 40 (CD40) gene (Figure 4G). CD40, a member of the tumor necrosis factor receptor (TNFR) superfamily, plays a crucial role in regulating immune responses42. Previous research indicated that receptors initiating NF-κB signaling are transcriptionally dysregulated in schizophrenia43, but this association had not been established in the context of a common variant affecting microglia-specific regulation of CD4044,45. These two examples demonstrate the capability of our model to not only identify causal genes but also, by offering cell-specific contexts, facilitate a better understanding of the mechanisms through which risk loci act.
Penetrance and pleiotropy of schizophrenia polygenic scores
Understanding the penetrance and pleiotropy of schizophrenia PRS is critical for predicting risk and guiding clinical interventions. We used a “leave-one-out” approach to create an independent PRS constructed from AFR-specific weights, and performed phenome-wide association studies (PheWAS) to explore its relationship with a range of mental and physical health problems.
Across 1,650 unique “phecodes”, PRS were robustly associated with schizophrenia (OR=1.21, 95% CI:[1.18,1.25]; P=6.31×10−44), bipolar disorder (OR=1.11, 95% CI:[1.09,1.14]; P=1.55×10−16), major depression (OR=1.05, 95% CI:[1.04,1.07]; P<3.89×10−14), alcohol-related (OR=1.05, 95% CI:[1.04,1.07]; P=1.10×10−12) and tobacco use disorders (OR=1.05, 95% CI:[1.03,1.06]; P=1.48×10−10), and suicidal attempt and ideation (OR=1.09, 95% CI:[1.06,1.12]; P=4.10×10-8) (Figure 5A; Supplementary Table 27). We also replicated previously reported associations 12,47 between schizophrenia PRS and increased risk of viral hepatitis (OR=1.06, 95% CI:[1.04,1.08]; P=4.10×10-8), upper respiratory infections (OR=1.04, 95% CI:[1.03,1.06]; P=1.77×10-8), and dental caries (OR=1.05, 95% CI:[1.03,1.06]; P=1.17×10−10), and lower risk of obstructive sleep apnea (OR=0.96, 95% CI:[0.95,0.98]; P=3.57×10-7) and hypertension (OR=0.96, 95% CI:[0.94,0.98]; P=4.88×10-6).
We extended this analysis to 49 common laboratory measurements, observing additional significant associations between PRS and increased levels of hematocrit (β=0.020, 95% CI:[0.014,0.026]; P=2.98×10−11), high density lipoprotein (HDL) cholesterol (β=0.017, 95% CI:[0.011,0.023]; P=1.50×10-7), and aspartate aminotransferase (AST) (β=0.013, 95% CI:[0.007,0.020]; P=6.69×10-5) (Extended Data Figure 7; Supplementary Table 28).
Overall, we found that findings largely mirrored corresponding results for EUR-trained PRS (Supplementary Tables 29-30). We directly compared results based on AFR- and EUR-trained PRS, observing a high degree of correlation (ρ=0.715; P=3.01×10−56) in effect sizes (Figure 5B; see Methods). Notably, the relationship between AFR scores and schizophrenia diagnosis (OR=1.21, 95% CI:[1.18,1.25]; P=6.31×10-44) was slightly attenuated compared to EUR scores (OR=1.32, 95% CI:[1.27,1.36]; P=1.14×10-55), likely reflecting the lower power of our available training data.
Similarly, direct comparison of effect sizes for PheWAS of AFR PRS in AFR versus EUR participants highlighted replicable cross-ancestry associations, and significant overall correlations (ρ=0.6; P=1.10×10−32) (Figure 5C; Supplementary Tables 31-32),
Among AFR replication cohorts, EHR data were uniquely available for AOU, presenting an opportunity to replicate our findings across healthcare systems. We found that higher PRS increased the odds of being diagnosed with schizophrenia (OR=1.21, 95% CI:[1.13,1.29]; P=1.45×10-7) and to a lesser degree, bipolar disorder (OR=1.06 95% CI:[1.01,1.12]; P=0.0329); substance addiction (OR=1.11, 95% CI:[1.07,1.15]; P=3.47×10-8), and tobacco use disorders (OR=1.04, 95% CI:[1.10,1.07]; P=5.30×10-5). We replicated our previous findings that higher PRS lowered the odds of obstructive sleep apnea (OR=0.93, 95% CI:[0.89,0.97]; P=0.0003) and hyperlipidemia (OR=0.95, 95% CI:0.92,0.98]; P=0.00385) (Supplementary Table 33). Effects estimated in AFR participants in MVP and AOU were significantly correlated (ρ=0.54; P=3.17×10−14; Figure 5D), albeit to a lesser degree than observed for AFR and EUR individuals in MVP. Comparing phenome-wide prevalences across datasets, we observed correlations of 0.94 between AFR and EUR participants, and 0.63 between AFR participants in MVP and AOU (Supplementary Note). This finding may reflect different lifetime exposures and patterns of healthcare utilization between military and civilian populations.
Discussion
This study significantly advances our understanding of schizophrenia’s genetic underpinnings by incorporating data for African ancestry populations. The integration of data from the CSP #572/MVP and AOU Research Programs with four case-control cohorts yielded a sample size of 13,012 affected individuals and 54,266 controls, and increases the sample size of an earlier 2009 study19 by a factor of 10, and our 201920 and 2021 studies7 by 2.2- and 1.75-fold, respectively. This is the first study to yield independently significant associations in African ancestry populations. Through detailed investigations of heritability, local ancestry, and cell-type enrichment, we underline the shared biological underpinnings of schizophrenia across populations, in spite of heterogeneous genetic effects and pervasive disparities in diagnosis.
The findings reveal both convergent and divergent aspects of the genetic architecture of schizophrenia across different ancestries. While many loci identified in European populations showed directionally consistent effects in individuals of African ancestry, our approach also highlighted loci with potentially population-specific effects. For example, GRIN2A harbors both rare coding48 and common variants that confer risk for schizophrenia24, with distinct associations between other non-coding variation, smoking behaviors49, and educational attainment50, results that more closely mirror the pattern of association observed in AFR populations.
Differences in allele frequency and patterns of LD have the potential to lower GWAS discovery power, but can also be leveraged via fine-mapping approaches to identify a common, credible set of putatively causal variants. This “trade-off” between power and resolution is best illustrated by the comparatively fewer number of independent SNPs detected in joint analyses of African and European ancestry datasets compared to EUR-EAS and expanded EUR-only analyses, contrasted with the larger reductions in credible set size in joint analyses with AFR. The fine-mapping results, particularly at loci such as CACNA1I, demonstrate how leveraging diverse genetic backgrounds can refine the localization of causal variants, potentially accelerating the identification of therapeutic targets.
Focusing on African-like populations uncovered population-specific associations upstream of PLXNA4, which encodes a molecule critical for axonal guidance51–53, and has been implicated in Alzheimer’s disease54, autism55, and innate inflammatory responses56. Another example is PMAIP1, which encodes the proapoptotic molecule, Noxa, a crucial determinant of cell fate in developing oligodendrocytes57 that has been shown to be upregulated in the blood of military service members with PTSD58 and in frontal brain of mice exposed to olanzapine59. We anticipate that further expansions in AFR sample sizes will yield further novel, biologically plausible insights, including allelic heterogeneity and pleiotropic effects. Comparable expansions of EUR sample sizes may yield diminishing returns in terms of locus discovery, as suggested by an independent EUR analysis of 6,579 cases and 155,490 controls from MVP, which recovered three well-established associations at ZNF804, LEMD2, and BCL11B.
Given the selected nature of the population utilizing VHA services, and higher rates of diagnosis among Black and African American individuals, we sought to characterize the common variant genetic architecture of schizophrenia. Directly measured h2 was virtually indistinguishable between veterans of African and European ancestry6,7, which supports the assertion that schizophrenia is no more or less “genetic” in one group versus another. Disparities in the rates of diagnosis could reflect the consequences of structural racism, both in the negative impact of exposure to adverse social conditions3, as well implicit biases that label schizophrenia as primarily a “Black disease”60,61. Notably, we observed that the stark differences in prevalence of schizophrenia diagnoses appear to have diminished in recent decades (Supplementary Note).
The investigation into cell-type-specific genetic architectures revealed significant enrichment of schizophrenia risk loci in excitatory and inhibitory neurons, consistent with previous findings in European populations24,30,31. Although aberrant myelination is known to play a role in neuropsychiatric disorders62 and the association between OPCs and schizophrenia is well-established30,31; however, the higher enrichment of OPCs in African ancestry individuals merits further investigation, as it may uncover novel biological pathways contributing to schizophrenia risk that are more pronounced in, or unique to, these populations.
Our exploration of the polygenic manifestations of currently indexable risk demonstrated resoundingly that AFR-derived scores are robustly transferable across cohorts, settings, and populations. Our PheWAS successfully recapitulated cross-trait associations with mental and physical health conditions, including both risk-increasing and apparently protective effects, and many which remained significant when limiting our analyses to individuals without a formal psychiatric diagnosis.
Despite nearly doubling the sample size of prior AFR GWAS7,20, we were still underpowered relative to contemporary EUR24 and EAS23 studies. Pointedly, our analysis is on a par with the discovery phase of a 2013 study that yielded genome-wide significant associations in 12 genomic regions63. It is important to note that the assembled study cohorts all originated in the United States, and therefore represent a fraction of the vast genomic (and experiential) diversity present in African diaspora populations. By definition, heritability is a population-specific phenomenon, and forthcoming research initiatives across continental Africa and the Caribbean64,65 have tremendous potential to accelerate genetic discovery and epidemiological insights. Lastly, though schizophrenia is a highly heritable disorder, there is considerable evidence for the role of social conditions, especially those related to structural racism2,3. Approaches that integrate social and biological risk factors are important for improving our understanding of the etiology of schizophrenia.
In conclusion, this study not only contributes valuable data to the field of psychiatric genetics but also highlights the critical importance of including diverse populations in genetic research. Large-scale genome sequencing endeavors should prioritize genetically diverse and otherwise underrepresented populations in order to fully realize their potential for novel discoveries64,66 and realize biological insights with equitable public health relevance to patients and communities worldwide, paving the way for advancements in precision medicine that are inclusive and beneficial for all populations.
Online Methods
Ethics/study approval
This study was approved by the VA Central Institutional Review Board (IRB), and participating studies received approval from their respective IRBs. All participants provided written informed consent.
Study Participants
Details of study ascertainment and assessment for the CSP/MVP, All of Us, COGS, GPC, PAARTNERS, and MGS are summarized in the Supplementary Note, and have otherwise been described elsewhere7,12,14–16,18–21,67,68
Genome-wide association studies, replication, and meta-analysis
We performed primary GWAS of schizophrenia in CSP/MVP, COGS, MGS, and PAARTNERS using imputed allelic dosages and logistic regression, as implemented in PLINK2, adjusting for age, sex, and 10 ancestry principal components (PCs).
For AOU, we performed logistic regression as implemented in HAIL using whole-genome sequencing data (release 7), and adjusting for age, sex, and 10 ancestry PCs.
We meta-analyzed AFR cohorts using inverse variance weighted meta-analysis (“fixed effects”), and combined results across ancestries using Han and Eskin’s random effects model with heterogeneity (RE2)69. Both models are implemented in the METASOFT software package.
Heritability estimation
We directly estimated heritability (SNP-h2) from imputed genotypes using genome-based restricted maximum likelihood (GREML)70 as implemented in the genome-wide complex trait analysis (GCTA) software70,71. We included the same GWAS covariates and used the [--grm-cutoff .05] flag to restrict analyses to approximately unrelated individuals. To enhance comparability of results, we downsampled the CSP/MVP data to include 6,000 cases and 18,000 screened controls of each ancestry (AFR and EUR). We estimated minor allele frequencies within each ancestry, and the absolute frequency difference (AFD)29, and partitioned SNPs into bins based on ad hoc MAF thresholds or AFD quintiles, followed by partitioned heritability analysis70. For ease of interpretability in the context of the published literature, we performed liability scale transformations assuming a 1% lifetime prevalence of schizophrenia.
Fine-mapping and credible sets
We used the Sum of Single Effects model (SuSiE)72,73 as implemented in the SuSiE-R package, to estimate causal credible SNP sets from GWAS meta-analysis results. First, we defined associated intervals around conditionally independent SNPs, collapsing all COJO variants within 250 kb windows. We calculated reference LD information for “super populations” in the 1000 Genomes Population Phase 3 data (i.e. EUR, AFR, EAS). Briefly, for each ancestral population represented in a given GWAS meta-analysis, we applied the following quality control steps: 1) we removed strand-ambiguous SNPs, multi-allelic SNPs, and SNPs with MAF less than 0.1%; 2) we retained only common SNPs across these ancestries and present the corresponding LD reference data. Next, we estimated credible sets at each associated interval, defined as the minimum number of ranked variants with cumulative posterior inclusion probability (PIP) greater than or equal to 0.95. We estimated credible sets based on EUR results alone, comparing these to corresponding sets derived from cross-ancestry GWAS findings. We evaluated improvements in fine-mapping on the basis of the size (the number of SNPs included in the set) and interval length (the total kb encompassed by each set). We used a probability threshold of 0.95 to define credible sets, and tentatively classified as causal any SNP with a posterior inclusion probability (PIP) >0.5.
We extended these analyses to the direct comparisons of ancestry-specific GWAS results using a recent extension of SuSiE, multi-ancestry sum of the single effects model (MESuSiE)74. Briefly, MESuSiE models ancestry-specific information for two populations (e.g., EUR-AFR, EUR-EAS, AFR-EAS) to enhance the precision and resolution of statistical fine-mapping. Like SuSiE-R, this method takes summary statistics as input, considers the varied LD patterns present in distinct ancestries, explicitly models shared and ancestry-specific causal effects, and employs a scalable variational inference algorithm for computation. SNPs in the resulting 95% credible sets had a non-zero effect in at least one population, and are classified as either “shared” or “ancestry-specific” depending on their modeled contribution to the credible set.
We also considered SuSiE-X75, another extension of the SuSiE framework, which has been shown to enhance precision of cross-population fine-mapping while maintaining well-calibrated false positive rates, and retaining the capability to identify population-specific causal variants. We followed the same steps as described above for processing ancestry-specific GWAS and LD information.
Local ancestry inference
We employed RFMix276 to deconvolve local ancestry among admixed AA individuals, implementing a local ancestry adjusted model analogous to the tractor method25. Specifically, we simultaneously modeled haploid dosages for AFR- and EUR-derived tracts, along with the number of AFR haplotypes carried by an individual at a locus (i.e. “hapcount”) using the local covariate functionality of PLINK277.
To better understand how admixture can impact the “portability” of schizophrenia findings, we compared conventional GWAS results for AFR and EUR (i.e. with global PC adjustment) with estimates based on deconvolved haplotypes in admixed AFR individuals. Across 270 replicated schizophrenia associations, 65% and 90% showed the same direction of effect in AFR (P=1×10-7) and EUR participants in MVP. When considering only AFR tracts in AFR participants, only 58% (P=0.007) of SNPs showed directionally consistent effects, compared to 69% (P=3.4×10−10) based on EUR tracts.
Cell-type enrichment at single cell resolution
We used the single-cell Disease-Relevance Scoring (scDRS)33 methodology to evaluate the aggregate expression of potential disease-associated genes obtained from EUR and AFR summary statistics utilizing Multi-marker Analysis of GenoMic Annotation (MAGMA)78. Each putative disease gene was weighted by its MAGMA Z-score from GWAS and inversely weighted by its gene-specific technical noise level in single-cell data obtained from an extensive dataset of 6.3 million nuclei generated from postmortem human dorsolateral prefrontal cortex of 1,494 donors. This process generated cell-specific raw disease scores. Additionally, we computed 1,000 sets of cell-specific raw control scores from matched control gene sets, ensuring similarity in gene set size, mean expression, and expression variance with the putative disease genes. Subsequently, we normalized both the raw disease scores and raw control scores for each cell, yielding normalized disease scores and normalized control scores. The computation of these scores utilized the default settings of the scDRS compute-score function. For downstream analysis, we conducted cell type-level assessments to link broad cell types to disease and explore heterogeneity in disease association across cells within each broad cell type. This was achieved using the scDRS perform-downstream() function with default settings. To address multiple testing, false discovery rate (FDR) was calculated using the Benjamini-Hochberg method.
TWAS concordance
To perform TWAS analysis, we utilized brain single nucleus (sn) transcriptomic imputation models from the PsychAD cohort. S-PrediXcan79 was used to perform ancestry-matched TWASs for schizophrenia GWASs. Consequently, TWAS z-scores across all cell-type/gene pairs were normalized (mean = 0; SD=1) within each ancestry to more easily compare results across ancestries. To assess concordance, only common cell-type/gene pairs were retained. Genes were ranked by meta-analysis p-value (performed on scaled z-scores). Correlation was calculated amongst all common genes in each cell type, and sequentially, genes were restricted to increasingly smaller sets (top 1,000 genes, 990 genes, etc.) to construct correlation trajectories. Only well-represented class-level cell types were retained for visualization.
Prediction of enhancer-gene interactions
To study E-P interactions affected by schizophrenia GWAS variants, we leveraged cell-type-specific E-P maps from a multi-omics dataset (joint snATAC-seq & snRNA-seq and cell-specific Hi-C) of developing brains using activity-by-contact (ABC) model36,37 (Extended Data Figure 6A). In line with authors’ instructions36,38, we filtered out E-P interactions that (i) had an ABC score < 0.015, (ii) involved ubiquitously expressed genes or genes on the Y chromosome, and (iii) included genes not expressed in major brain cell types. To further refine our focus to schizophrenia GWAS, we decided to keep only E-P links (i) overlapping conditionally independent SNPs (peaks participating in E-P links were extended by 100bp on both sides to increase the overlap), (ii) LD buddies of conditionally independent SNPs (R2 ≥ 0.8), or (iii) SNPs within 95% credible set and PPI > 0.5 (model AFR+EUR+EAS), resulting in 34,436 E-P links. To limit the number of causal genes per conditionally independent SNPs, we applied the ABC-MAX approach38, allowing only one E-P per conditionally independent SNP, resulting in 1,538 E-P links (Extended Data Figure 6B).
PRS Profiling
We constructed PGS from published and current GWAS results (the “training” datasets), testing individual-level scores for association with case-control status in MVP or replication cohorts (the “target” dataset). We applied a recently-developed Bayesian framework that applies continuous shrinkage to test statistics, PRS-CSx27, using HapMap 3 SNPs and 1000 Genomes Project LD reference panel (EUR and AFR). We also constructed scores based on genome-wide significant SNPs, following LD-based clumping in the appropriate KGP3 population (r2>0.1; 500kb window). Scores were constructed by summing the number of tested alleles weighted by their effect estimates (e.g., the log of the allelic odds ratio). We tested for case-control differences by logistic regression using age, sex, and the first six ancestry principal components (PCs).
PheWAS and LabWAS
We employed phenome-wide association studies (PheWAS) to explore the relationships between neuropsychiatric PRSs and “phecodes” representing groupings of related ICD-9/10 billing codes80.When testing individual phecodes, we required cases and controls to have ≥2 and zero codes, respectively. We applied logistic regression to test scaled PRSs (mean=0; SD=1) for association with phecodes within ancestry groups, covarying for age, age2, gender, and six ancestry principal components (PCs). We performed LabWAS across mean, median, maximum, and minimum normalized values in individuals with at least two observed measurements using linear regression and the same covariates as in PheWAS analyses.
We performed a series of sensitivity analyses, co-varying for selected diagnoses or treatment with antipsychotics, mood-stabilizers, and antidepressants, or removing individuals with any lifetime diagnosis of psychotic, mood, or substance disorders.
We compared effect sizes across PheWAS results sets using an pseudo-independent set of phecodes, based on the most significant among ontologically nested terms. We calculated Pearson’s correlations based on standardized effect sizes (i.e., Z-scores).
Reporting summary
Additional information about our study design is available in the Nature Research Reporting Summary linked to this paper.
Data availability
All data produced in the present study are available upon reasonable request to the authors
Code availability
We utilized second generation PLINK for primary quality control and descriptive analyses (https://www.cog-genomics.org/plink2), For haplotype phasing and genotype imputation, we used SHAPEIT4 (https://odelaneau.github.io/shapeit4/) and Minimac4 (https://github.com/statgen/Minimac4), respectively. For local ancestry inference and defining ancestry tracts, we utilized RFMix2 (https://github.com/slowkoni/rfmix). Conditional and joint analysis (COJO) and heritability estimation utilized GCTA (https://cnsgenomics.com/software/gcta). Meta-analyses were conducted using METASOFT (https://www.zarlab.xyz/tag/metasoft/). For statistical fine mapping, we used SuSiE-R (https://github.com/stephenslab/susieR), SuSiE-X (https://github.com/getian107/SuSiEx), and MESuSiE (https://github.com/borangao/MESuSiE). We estimated gene-level test statistics using MAGMA (https://cncr.nl/research/magma/) and tested for cell-type specific enrichment using scDRS (https://github.com/martinjzhang/scDRS). We used S-PrediXcan (https://github.com/hakyimlab/MetaXcan) to directly impute TWAS results from GWAS summary statistics. To characterize enhancer-gene links, we adopted the ABC-MAX framework (https://github.com/EngreitzLab/ABC-Max-pipeline). For PRS, we used PRS-CSx to integrate GWAS summary statistics from multiple populations and estimate posterior SNP effects via continuous shrinkage (https://github.com/getian107/PRScsx).
EXTENDED TABLES AND FIGURES
CONSORTIA AUTHORS
Cooperative Studies Program (CSP) #572
Mihaela Aslan PC,EC,1, M. Antonelli PC, John Concato PC,EC,1, F. Cunningham PC, M. Gaziano PC,EC,2, Theresa Gleason PC,3, Philip D. Harvey PC,EC,4, Grant D. Huang PC,3, Thomas R. Kosten PC,EC,6, Alan C. Swann 6, P. Miller PC,1, Timothy J. O’Leary PC,3, Ronald Przygodzki PC,3, Pamela Sklar PC,EC,5, Hongyu Zhao PC,EC,1, Ayman Fanous EC,20,7, Shrikant Mane 1, M. Corsey 5, L. Zaluda 5, J. Johnson 4, D. Cavaliere 11, V. Jeanpaul 1, Alysia Maffucci 1, L. Mancini 1, P. Peduzzi PC,1, J. Deen 2, G. Muldoon 2, S. Whitbourne 2, L. Adamson 8, L. Calais 8, G. Fuldauer 8, R. Kushner 8, G. Toney 8, M. Lackey 8, A. Mank 8, N. Mahdavi 8, G. Villarreal 8, E. C. Muly 9, F. Amin 9, M. Dent 9, J. Wold 9, B. Fischer 10, A. Elliott 10, C. Felix 10, G. Gill 10, P. E. Parker 11, C. Logan 11, J. McAlpine 11, Lynn E. DeLisi 12, S. G. Reece 12, M. B. Hammer 12, M. de Asis PC,12, D. Agbor-Tabie 13, W. Goodson 13, M.S. Bauer PC,14, Mary Brophy PC,EC,14, M. Aslam 14, M. Grainger 14, Neil Richtand 14, Alexander Rybalsky 14, R. Al Jurdi 6, E. Boeckman 6, T. Natividad 6, D. Smith 6, M. Stewart 6, S. Torres 6, Z. Zhao 6, A. Mayeda 15, A. Green 15, J. Hofstetter 15, S. Ngombu 15, M. K. Scott 15, A. Strasburger 15, J. Sumner 15, G. Paschall 16, J. Mucciarelli 16, R. Owen 16, S. Theus 16, D. Tompkins 16, S. G. Potkin 17, C. Reist 17, M. Novin 17, S. Khalaghizadeh 17, R. Douyon 4, N. Kumar 4, B. Martinez 4, S. R. Sponheim 18, T. L. Bender 18, H. L. Lucas 18, A. M. Lyon 18, M. P. Marggraf 18, L. H. Sorensen 18, C. R. Surerus 18, C. Sison 19, J. Amato 19, D. R. Johnson 19, N. Pagan-Howard 19, L. A. Adler 20, S. Alperin 20, T. Leon 20,Tim B. Bigdeli 20,EC, Peter B. Barr20, Zoe E. Neale 20, Sundar Natarajan 20, K. M. Mattocks 21, N. Araeva 21, J. C. Sullivan 21, T. Suppes 22, K. Bratcher 22, L. Drag 22, E.G. Fischer 22, L. Fujitani 22, S. Gill 22, D. Grimm 22, J. Hoblyn 22, T. Nguyen 22, E. Nikolaev 22, L. Shere 22, R. Relova 22, A. Vicencio 22, M. Yip 22, I. Hurford 23, S. Acheampong 23, G. Carfagno 23, G.L. Haas 24, C. Appelt 24, E. Brown 24, B. Chakraborty 24, E. Kelly 24, G. Klima 24, S. Steinhauer 24, R. A. Hurley 25, R. Belle 25, D. Eknoyan 25, K. Johnson 25, J. Lamotte 25, E. Granholm 26, K. Bradshaw 26, J. Holden 26, R. H. Jones 26, T. Le 26, I. G. Molina 26, M. Peyton 26, I. Ruiz 26, L. Sally 26, J. B. Lohr PC,26, T. Patterson PC,26, A. Tapp 27, S. Devroy 27, V. Jain 27, N. Kilzieh 27, L. Maus 27, K. Miller 27, H. Pope 27, A. Wood 27, E. Meyer 28, P. Givens 28, P. B. Hicks 28, S. Justice 28, K. McNair 28, J. L. Pena 28, D. F. Tharp 28, L. Davis 29, M. Ban 29, L. Cheatum 29, P. Darr 29, W. Grayson 29, J. Munford 29, B. Whitfield 29, E. Wilson 29, S. E. Melnikoff 7, B. L. Schwartz 7, M. A. Tureson 7, D. D’Souza 1, K. Forselius 1, John Ko1, John Russo1, M. Ranganathan 1, L. Rispoli 1, Yuli Li1, Frederick Sayward1, Nallakkandi Rajeevan1, Krishnan Radhakrishnan1, Kei-Hoi Cheung1, M. Sather 8, C. Colling 8, C. Haakenson 8, D. Krueger 8, Sumitra Muralidhar 3, J. Kelsoe PC,26, W. Farwell EC,2, Anil Malhotra EC,30, Polly Palacios EC,31, Thomas Lehner PC,32, Steven R. Marder PC,33, R. Freedman PC,34, Stephen Strakowski PC,35
EC Executive Committee; PC Planning Committee; 1 Clinical Epidemiology Research Center (CERC), VA Connecticut Healthcare System, West Haven, CT; 2 Massachusetts Area Veterans Epidemiology, Research, and Information Center (MAVERIC), Jamaica Plain, MA 3 Office of Research and Development, Veterans Health Administration, Washington, DC; 4 Bruce W. Carter Miami Veterans Affairs (VA) Medical Center, Miami, FL; 5 Mental Illness Research, Education and Clinical Center VISN2, James J. Peters VA Medical Center, Bronx, NY; 6 Michael E. DeBakey VA Medical Center, Houston, TX 7 Washington DC VA Medical Center, Washington, DC; 8 Raymond G. Murphy VA Medical Center, Albuquerque, NM; 9 Atlanta VA Health Care System, Decatur, GA; 10 VA Maryland Health Care System, Baltimore, MD; 11 Birmingham VA Medical Center, Birmingham, AL; 12 VA Boston Healthcare System, Brockton, MA; 13 Ralph H. Johnson VA Medical Center, Charleston, SC; 14 Cincinnati VA Medical Center, Cincinnati, OH; 15 Richard L. Roudebush VA Medical Center, Indianapolis, IN; 16 John L. McClellan Memorial Veterans Hospital, Little Rock, AR; 17 VA Long Beach Healthcare System, Long Beach, CA; 18 Minneapolis VA Health Care System, Minneapolis, MN; 19 VA Hudson Valley Health Care System, Montrose, NY; 20 VA New York Harbor Healthcare System, Brooklyn, NY; 21 VA Central Western Massachusetts Healthcare System, Leeds, MA; 22 VA Palo Alto Health Care System, Palo Alto, CA; 23 Corporal Michael J. Crescenz VA Medical Center, Philadelphia, PA; 24 VA Pittsburgh Healthcare System, Pittsburgh, PA; 25 W.G. (Bill) Hefner VA Medical Center, Salisbury, NC; 26 VA San Diego Healthcare System, San Diego, CA; 27 VA Puget Sound Health Care System, Tacoma, WA; 28 Central Texas Veterans Health Care System, Temple, TX; 29 Tuscaloosa VA Medical Center, Tuscaloosa, AL. 30 Department of Psychiatry, Zucker Hillside Hospital, Donald and Barbara Zucker School of Medicine at Hofstra/Northwell, Glen Oaks, NY; 31 VISN 8, VA Sunshine Healthcare Network, James A. Haley Veterans’ Hospital, Tampa, FL; 32 National Institute of Mental Health, Bethesda, MD; 33 VA Greater Los Angeles Healthcare System, Los Angeles, CA; 34 Department of Psychiatry, University of Colorado School of Medicine, Anschutz Medical Campus,Aurora, CO; 35 Department of Psychiatry, University of Texas Dell Medical School, Austin, TX
Consortium on the Genetics of Schizophrenia (COGS) Investigators
David L Braff, MD 1,2, Monica E Calkins, PhD 3, Robert R Freedman, MD 4, Michael F Green, PhD 5,6,7, Tiffany A Greenwood, PhD 1, Raquel E Gur, MD, PhD 3,8-9, Ruben C Gur, PhD 3,8-9, Laura C Lazzeroni, PhD 10,11, Gregory A Light, PhD 1,2, Keith H Nuechterlein, PhD 6,7, Allen D Radant, MD 12,13, Larry J Seidman, PhD 14,15, Larry J Siever, MD 16,17, Jeremy M Silverman, PhD 16,17, William S Stone, PhD 14,15, Catherine A Sugar, PhD 18, Neal R Swerdlow, MD, PhD 1, Debby W Tsuang, MD 12,13, Ming T Tsuang, MD, PhD, DSc 1,19,20, Bruce I Turetsky, MD 3
1 Department of Psychiatry, University of California, La Jolla, San Diego, CA, USA; 2 VISN-22 Mental Illness, Research, Education and Clinical Center (MIRECC), VA San Diego Healthcare System, San Diego, CA, USA; 3 Department of Psychiatry, University of Pennsylvania Perelman School of Medicine, Philadelphia, PA, USA; 4 Department of Psychiatry, University of Colorado Denver, Aurora, CO, USA; 5 VA Greater Los Angeles Healthcare System, Los Angeles, CA, USA; 6 Department of Psychiatry and Biobehavioral Sciences and 7 Semel Institute for Neuroscience and Human Behavior, Geffen School of Medicine, University of California Los Angeles, Los Angeles, CA, USA; 8 Department of Child & Adolescent Psychiatry, University of Pennsylvania Perelman School of Medicine; 9 Lifespan Brain Institute, Children’s Hospital of Philadelphia, Philadelphia, PA, USA; 10 Departments of Psychiatry and Behavioral Sciences and 11 Pediatrics, Stanford University, Stanford, CA, USA; 12 Department of Psychiatry and Behavioral Sciences, University of Washington, Seattle, WA, USA; 13 VA Puget Sound Health Care System, Seattle, WA, USA; 14 Department of Psychiatry, Harvard Medical School, Boston, MA, USA; 15 Massachusetts Mental Health Center Public Psychiatry Division of the Beth Israel Deaconess Medical Center, Boston, MA, USA; 16 Department of Psychiatry, Icahn School of Medicine at Mount Sinai, NY, USA; 17 James J. Peters Veterans Affairs Medical Center, Bronx, NY, USA; 18 Department of Biostatistics, University of California Los Angeles School of Public Health, Los Angeles, CA, USA; 19 Institute for Genomic Medicine, University of California San Diego, La Jolla, CA, USA; 20 Harvard Institute of Psychiatric Epidemiology and Genetics, Boston, MA, USA.
Genomic Psychiatry Cohort (GPC) Investigators
Michele T Pato MD 1,2, Carlos N Pato, MD, PhD 1,2, Tim B Bigdeli, PhD 1,2,3, Ayman H Fanous, MD 1,2,3, Steven A McCarroll, PhD 4,5, Peter F Buckley, MD 6, Mark J. Daly 7,8,9,5, James A Knowles MD, PhD 2,10, Douglas S Lehrer, MD 11, Dolores Malaspina, MD, MSPH 12,13, Mark H Rapaport, MD 14, Jeffrey J Rakofsky, MD 14, Janet L Sobell, PhD 15, Giulio Genovese, PhD 4,5, Penelope Georgakopoulos, DrPH 2, Jacquelyn L Meyers, PhD 1, Roseann E Peterson, PhD 6, Helena Medeiros, MSW 2, Jorge Valderrama, PhD 1,2, Eric D Achtyes, MD 16, Roman Kotov, PhD 17, Colony Abbott, MPH 16, Maria Helena Azevedo, PhD 18, Richard A Belliveau, Jr, BA 4, Elizabeth Bevilacqua, BS 19, Evelyn J Bromet, PhD 17, William Byerley, MD 20, Celia Barreto Carvalho, PhD 21, Sinéad B Chapman, MS 4, Lynn E DeLisi, MD 22,23, Ashley L Dumont, BASc 4, Colm O’Dushlaine, PhD 4, Oleg V Evgrafov, PhD 2,10, Laura J Fochtmann, MD 17, Diane Gage 4, James L Kennedy, MD 24, Becky Kinkead, PhD 14, Antonio Macedo, PhD 18, Jennifer L Moran, PhD 4, Christopher P Morley, PhD 25-27, Mantosh J Dewan, MD 27, James Nemesh 4, Diana O Perkins, MD, MPH 28, Shaun M Purcell, PhD 4,29, Edward M Scolnick, MD 4, Brooke M Sklar, MA 15, Pamela Sklar, MD, PhD 12,13, Jordan W Smoller, MD, ScD 4,23,30,31, Patrick F Sullivan, MD, FRANZCP 28,32, Humberto Nicolini, MD 33, Conrad O Iyegbe, PhD 34, Fabio Macciardi, MD, PhD 35, Stephen R Marder, MD 36,37, Michael A Escamilla, MD 38, Ruben C Gur, PhD 39-41, Raquel E Gur, MD, PhD 39-41, Tiffany A Greenwood, PhD 42, David L Braff, MD 42,43, Marquis P Vawter, PhD, MA, MS 35
1 Department of Psychiatry and Behavioral Sciences and 2 Institute for Genomic Health, SUNY Downstate Medical Center, Brooklyn, NY, USA; 3 VA New York Harbor Healthcare System, Brooklyn, NY, USA; 4 Stanley Center for Psychiatric Research, Broad Institute of MIT and Harvard, Cambridge, MA, USA; 5 Department of Genetics, Harvard Medical School, Boston, MA, USA; 6 School of Medicine, Virginia Commonwealth University, Richmond, VA, USA; 7 Institute for Molecular Medicine Finland (FIMM), University of Helsinki, Helsinki, Finland; 8 Analytic and Translational Genetics Unit, Massachusetts General Hospital, Boston, MA, USA; 9 Program in Medical and Population Genetics, Broad Institute of Harvard and MIT, Cambridge, MA, USA; 10 Department of Cell Biology, SUNY Downstate Medical Center, Brooklyn, NY, USA; 11 Department of Psychiatry, Wright State University, Dayton, OH, USA; 12 Departments of Psychiatry and 13 Genetics & Genomics, Icahn School of Medicine at Mount Sinai, NY, USA; 14 Department of Psychiatry and Behavioral Sciences, Emory University, Atlanta, GA, USA; 15 Department of Psychiatry & Behavioral Sciences, University of Southern California, Los Angeles, CA, USA; 16 Cherry Health and Michigan State University College of Human Medicine, Grand Rapids, MI, USA; 17 Department of Psychiatry, Stony Brook University, Stony Brook, NY, USA; 18 Institute of Medical Psychology, Faculty of Medicine, University of Coimbra, Coimbra, PT; 19 Beacon Health Options, Boston, MA, USA; 20 Department of Psychiatry, University of California, San Francisco, CA, USA; 21 Faculty of Social and Human Sciences, University of Azores, PT; 22 VA Boston Healthcare System, Brockton, MA, USA; 23 Department of Psychiatry, Harvard Medical School, Boston, MA, USA; 24 Neurogenetics Laboratory, Campbell Family Mental Health Research Institute, Centre for Addiction and Mental Health; Department of Psychiatry, University of Toronto, ON, CA; 25 Departments of Public Health and Preventive Medicine, 26 Family Medicine, and 27 Psychiatry and Behavioral Sciences, State University of New York, Upstate Medical University, Syracuse, NY, USA; 28 Department of Psychiatry, University of North Carolina, Chapel Hill, NC, USA; 29 Department of Psychiatry, Brigham and Women’s Hospital, Boston, MA, USA; 30 Department of Psychiatry, Massachusetts General Hospital, Boston, MA, USA; 31 Department of Epidemiology, Harvard T.H. Chan School of Public Health, Boston, MA, USA; 32 Medical Epidemiology and Biostatistics, Karolinska Institutet, Solna, SE; 33 Carracci Medical Group, Mexico City, MX; 34 Department of Psychosis Studies, King’s College London, London, UK; 35 Department of Psychiatry and Human Behavior, University of California, Irvine, CA, USA; 36 Department of Psychiatry and Biobehavioral Sciences and 37 Semel Institute for Neuroscience and Human Behavior, Geffen School of Medicine, University of California Los Angeles, Los Angeles, CA, USA; 38 Department of Psychiatry, University of Texas Rio Grande Valley School of Medicine; 39 Departments of Psychiatry and 40 Child & Adolescent Psychiatry and 41 Lifespan Brain Institute, University of Pennsylvania Perelman School of Medicine and Children’s Hospital of Philadelphia, Philadelphia, PA, USA; 42 Department of Psychiatry, University of California, La Jolla, San Diego, CA, USA; 43 VISN-22 Mental Illness, Research, Education and Clinical Center (MIRECC), VA San Diego Healthcare System, San Diego, CA, USA
Project Among African-Americans to Explore Risks for Schizophrenia (PAARTNERS)
Rodney CP Go, PhD 1, Raquel Gur, MD, PhD 2, Ruben Gur, PhD 2, Vish Nimgaonkar, MD, PhD 3, Bernive Devlin, PhD 3, Monica Calkins, PhD 2, Christian Kohler, MD 2, Roberta May 1; Charlie Swanson, Jr., MD 1, Laura Montgomery-Barefield, MD 1, Tolulope Aduroja, MD 1, Ryan Coleman 1; Rakesha Garner 1; Lee Prichett, RN 1; Thomas Kelley, RN 1; Marguerite Ryan Dickson, MS 1, Joseph P. McEvoy, MD 4, Linda Blalock, RN 4, Karen Richardson, MS 5, Deirdre Evans-Cosby, MD 6, George W. Woods, MD 6, Kendaly Meadows, RN 6, Sandra Cummings MSW. 6, Cara Stephens LCSW 6, Kent Baker 6, Shirley Hendrix 7, Cynthia Gilliard 7, Wanda Smalls-Smith 7, Steven McLeod-Bryant 7, Joel Wood, BS 3, Mary Miller, L.P.N. 3, Frank Fleischer, M.B.A. 3, Kristin Beizai, MD 8, Marie Tobin, MD 8, Alyssa English, MD 8, Richard Sanders, BS 8, Shelia A. Dempsey, ADN 8, Martha Velez, CSP 8, Marianne Smith, BS, MA 8, Martha Garriott, MS, NCC 8, Nancy Fowler 8, Derrick W. Allen, MSSW 8, Phyllis Meyer, BS, PA 8, Lynn Heustess, BS 8
1 University of Alabama at Birmingham, Birmingham, AL, USA; 2 Department of Child & Adolescent Psychiatry, University of Pennsylvania Perelman School of Medicine, Philadelphia, PA, USA; 3 Department of Psychiatry, University of Pittsburgh, Pittsburgh, Pennsylvania, USA; 4 Duke University, Durham, NC, USA; 5 University of Mississippi, University, MS, USA; 6 Morehouse School of Medicine, Atlanta, GA, USA; 7 Medical University of South Carolina, Charleston, SC, USA; 8 University of Tennessee, Knoxville, TN, USA
PsychAD Consortium
Aram Hong (1, 4, 6, 7); Athan Z. Li (10, 12); Biao Zeng (1, 4, 6, 7); Chenfeng He (9, 12); Chirag Gupta (9, 12); Christian Porras (1, 4, 6, 7); Clara Casey (1, 4, 6, 7); Colleen A. McClung (18); Collin Spencer (1, 4, 6, 7); Daifeng Wang (9, 10, 12); David A. Bennett (19); David Burstein (1, 2, 4, 6, 7, 8); Deepika Mathur (1, 4, 6, 7); Donghoon Lee (1, 4, 6, 7); Fotios Tsetsos (1, 2, 4, 6, 7); Gabriel E. Hoffman (1, 2, 4, 6, 7, 8); Genadi Ryan (13, 17); Georgios Voloudakis (1, 2, 3, 4, 6, 7, 8); Hui Yang (1, 4, 6, 7); Jaroslav Bendl (1, 4, 6, 7); Jerome J. Choi (11, 12); John F. Fullard (1, 4, 6, 7); Kalpana H. Arachchilage (9, 12); Karen Therrien (1, 4, 6, 7); Kiran Girdhar (1, 4, 6, 7); Lars J. Jensen (21); Lisa L. Barnes (19); Logan C. Dumitrescu (22, 23); Lyra Sheu (1, 4, 6, 7); Madeline R. Scott (18); Marcela Alvia (1, 4, 6, 7); Marios Anyfantakis (1, 4, 6, 7); Maxim Signaevsky (6, 7); Mikaela Koutrouli (1, 4, 6, 7, 21); Milos Pjanic (1, 4, 6, 7); Monika Ahirwar (13, 17); Nicolas Y. Masse (1, 4, 6, 7); Noah Cohen Kalafut (10, 12); Panos Roussos (1, 2, 4, 6, 7, 8); Pavan K. Auluck (20); Pavel Katsel (6); Pengfei Dong (1, 4, 6, 7); Pramod B. Chandrashekar (9, 12); Prashant N.M. (1, 4, 6, 7); Rachel Bercovitch (1, 4, 6, 7); Roman Kosoy (1, 4, 6, 7); Sanan Venkatesh (1, 2, 4, 6, 7); Saniya Khullar (9, 12); Sayali A. Alatkar (10, 12); Seon Kinrot (1, 4, 6, 7); Stathis Argyriou (1, 4, 6, 7); Stefano Marenco (20); Steven Finkbeiner (13, 14, 15, 16, 17); Steven P. Kleopoulos (1, 4, 6, 7); Tereza Clarence (1, 4, 6, 7); Timothy J. Hohman (22, 23); Ting Jin (9, 12); Vahram Haroutunian (5, 6, 7, 8); Vivek G. Ramaswamy (13, 17); Xiang Huang (12); Xinyi Wang (1, 4, 6, 7); Zhenyi Wu (1, 4, 6, 7); Zhiping Shao (1, 4, 6, 7)
PsychAD Consortium Affiliations
1: Center for Disease Neurogenomics, Icahn School of Medicine at Mount Sinai, New York, NY, USA
2: Center for Precision Medicine and Translational Therapeutics, James J. Peters VA Medical Center, Bronx, NY, USA
3: Department of Artificial Intelligence and Human Health, Icahn School of Medicine at Mount Sinai, New York, NY, USA
4: Department of Genetics and Genomic Sciences, Icahn School of Medicine at Mount Sinai, New York, NY, USA
5: Department of Neuroscience, Icahn School of Medicine at Mount Sinai, New York, NY, USA
6: Department of Psychiatry, Icahn School of Medicine at Mount Sinai, New York, NY, USA
7: Friedman Brain Institute, Icahn School of Medicine at Mount Sinai, New York, NY, USA
8: Mental Illness Research, Education and Clinical Center VISN2, James J. Peters VA Medical Center, Bronx, NY, USA
9: Department of Biostatistics and Medical Informatics, University of Wisconsin-Madison, Madison, WI, USA
10: Department of Computer Sciences, University of Wisconsin-Madison, Madison, WI, USA
11: Department of Population Health Sciences, University of Wisconsin-Madison, Madison, WI, USA
12: Waisman Center, University of Wisconsin-Madison, Madison, WI, USA
13: Center for Systems and Therapeutics, Gladstone Institutes, San Francisco, CA, USA
14: Department of Neurology, University of California San Francisco, San Francisco, CA, USA
15: Department of Physiology, University of California San Francisco, San Francisco, CA, USA
16: Neuroscience and Biomedical Sciences Graduate Programs, University of California San Francisco, San Francisco, CA, USA
17: Taube/Koret Center for Neurodegenerative Disease Research, Gladstone Institutes, San Francisco, CA, USA
18: Department of Psychiatry, University of Pittsburgh School of Medicine, Pittsburgh, PA, USA
19: Rush Alzheimer’s Disease Center and Department of Neurological Sciences, Rush University Medical Center, Chicago, IL, USA
20: Human Brain Collection Core, National Institute of Mental Health-Intramural Research Program, Bethesda, MD, USA
21: Novo Nordisk Foundation Center for Protein Research, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, Denmark
22: Vanderbilt Genetics Institute, Vanderbilt University Medical Center, Nashville, TN, USA
23: Vanderbilt Memory & Alzheimer’s Center, Vanderbilt University Medical Center, Nashville, TN, USA
ACKNOWLEDGEMENTS
This research was supported by the Department of Veterans Affairs Cooperative Studies Program (CSP) #572 and the Million Veteran Program (MVP-000, MVP-006 and MVP-076). The Million Veteran Program is supported by the Office of Research and Development, Department of Veterans Affairs. The contents do not represent the views of the U.S. Department of Veterans Affairs or the United States Government. Complete acknowledgements for CSP #572 and MVP are given in the Supplementary Materials.
This study was supported by the Veterans Affairs Merit grants: BX004189 (to P.R.). This work was supported by the National Institutes of Health (NIH): R01MH125246 (to P.R. and T.B.), K08MH122911 (to G.V.), R01AG078657 (to G.V.), R01MH104964 (to T.B., A.H.F., M.T.P., C.N.P.), R01AG067025 (to P.R.), U01MH116442 (to P.R.), and U24AG087563 (to P.R.).
The GPC was supported by grants R01 MH085548 and R01 MH104964 from the National Institute of Mental Health (NIMH), and genotyping of samples was provided by the Stanley Center for Psychiatric Research at Broad Institute.
COGS was supported by grants R01 MH065571, R01 MH065588, R01 MH065562, R01 MH065707, R01 MH065554, R01 MH065578, R01 MH065558, R01 MH86135, and R01 MH094320 from the National Institute of Mental Health.
Funding support for the Genome-Wide Association of Schizophrenia Study was provided by the NIMH (R01 MH67257, R01 MH59588, R01 MH59571, R01 MH59565, R01 MH59587, R01 MH60870, R01 MH59566, R01 MH59586, R01 MH61675, R01 MH60879, R01 MH81800, U01 MH46276, U01 MH46289, U01 MH46318, U01 MH79469, and U01 MH79470) and the genotyping of samples was provided through the Genetic Association Information Network (GAIN).
Dr. Harvey has served as a consultant to multiple pharmaceutical companies and device manufacturers on phase 2 or 3 development; this consulting work has been determined to be unrelated to the content of the paper. No other authors report any relevant conflicts of interest.
Footnotes
REFERENCES
- 1.↵
- 2.↵
- 3.↵
- 4.↵
- 5.↵
- 6.↵
- 7.↵
- 8.↵
- 9.↵
- 10.
- 11.↵
- 12.↵
- 13.↵
- 14.↵
- 15.↵
- 16.↵
- 17.↵
- 18.↵
- 19.↵
- 20.↵
- 21.↵
- 22.↵
- 23.↵
- 24.↵
- 25.↵
- 26.↵
- 27.↵
- 28.↵
- 29.↵
- 30.↵
- 31.↵
- 32.↵
- 33.↵
- 34.↵
- 35.↵
- 36.↵
- 37.↵
- 38.↵
- 39.↵
- 40.↵
- 41.↵
- 42.↵
- 43.↵
- 44.↵
- 45.↵
- 46.
- 47.↵
- 48.↵
- 49.↵
- 50.↵
- 51.↵
- 52.
- 53.↵
- 54.↵
- 55.↵
- 56.↵
- 57.↵
- 58.↵
- 59.↵
- 60.↵
- 61.↵
- 62.↵
- 63.↵
- 64.↵
- 65.↵
- 66.↵
- 67.↵
- 68.↵
- 69.↵
- 70.↵
- 71.↵
- 72.↵
- 73.↵
- 74.↵
- 75.↵
- 76.↵
- 77.↵
- 78.↵
- 79.↵
- 80.↵