Quantitative trait loci mapping of gene expression and chromatin accessibility in primary fibroblast reveals shared allelic effects between Latin American and European ancestries

Quantitative trait loci mapping of gene expression and chromatin accessibility in primary fibroblast reveals shared allelic effects between Latin American and European ancestries | bioRxiv /* */ /* */ <!-- <!-- /*! * yepnope1.5.4 * (c) WTFPL, GPLv2 */ (function(a,b,c){function d(a){return"[object Function]"==o.call(a)}function e(a){return"string"==typeof a}function f(){}function g(a){return!a||"loaded"==a||"complete"==a||"uninitialized"==a}function h(){var a=p.shift();q=1,a?a.t?m(function(){("c"==a.t?B.injectCss:B.injectJs)(a.s,0,a.a,a.x,a.e,1)},0):(a(),h()):q=0}function i(a,c,d,e,f,i,j){function k(b){if(!o&&g(l.readyState)&&(u.r=o=1,!q&&h(),l.onload=l.onreadystatechange=null,b)){"img"!=a&&m(function(){t.removeChild(l)},50);for(var d in y[c])y[c].hasOwnProperty(d)&&y[c][d].onload()}}var j=j||B.errorTimeout,l=b.createElement(a),o=0,r=0,u={t:d,s:c,e:f,a:i,x:j};1===y[c]&&(r=1,y[c]=[]),"object"==a?l.data=c:(l.src=c,l.type=a),l.width=l.height="0",l.onerror=l.onload=l.onreadystatechange=function(){k.call(this,r)},p.splice(e,0,u),"img"!=a&&(r||2===y[c]?(t.insertBefore(l,s?null:n),m(k,j)):y[c].push(l))}function j(a,b,c,d,f){return q=0,b=b||"j",e(a)?i("c"==b?v:u,a,b,this.i++,c,d,f):(p.splice(this.i++,0,a),1==p.length&&h()),this}function k(){var a=B;return a.loader={load:j,i:0},a}var l=b.documentElement,m=a.setTimeout,n=b.getElementsByTagName("script")[0],o={}.toString,p=[],q=0,r="MozAppearance"in l.style,s=r&&!!b.createRange().compareNode,t=s?l:n.parentNode,l=a.opera&&"[object Opera]"==o.call(a.opera),l=!!b.attachEvent&&!l,u=r?"object":l?"script":"img",v=l?"script":u,w=Array.isArray||function(a){return"[object Array]"==o.call(a)},x=[],y={},z={timeout:function(a,b){return b.length&&(a.timeout=b[0]),a}},A,B;B=function(a){function b(a){var a=a.split("!"),b=x.length,c=a.pop(),d=a.length,c={url:c,origUrl:c,prefixes:a},e,f,g;for(f=0;f<d;f++)g=a[f].split("="),(e=z[g.shift()])&&(c=e(c,g));for(f=0;f<b;f++)c=x[f](c);return c}function g(a,e,f,g,h){var i=b(a),j=i.autoCallback;i.url.split(".").pop().split("?").shift(),i.bypass||(e&&(e=d(e)?e:e[a]||e[g]||e[a.split("/").pop().split("?")[0]]),i.instead?i.instead(a,e,f,g,h):(y[i.url]?i.noexec=!0:y[i.url]=1,f.load(i.url,i.forceCSS||!i.forceJS&&"css"==i.url.split(".").pop().split("?").shift()?"c":c,i.noexec,i.attrs,i.timeout),(d(e)||d(j))&&f.load(function(){k(),e&&e(i.origUrl,h,g),j&&j(i.origUrl,h,g),y[i.url]=2})))}function h(a,b){function c(a,c){if(a){if(e(a))c||(j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}),g(a,j,b,0,h);else if(Object(a)===a)for(n in m=function(){var b=0,c;for(c in a)a.hasOwnProperty(c)&&b++;return b}(),a)a.hasOwnProperty(n)&&(!c&&!--m&&(d(j)?j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}:j[n]=function(a){return function(){var b=[].slice.call(arguments);a&&a.apply(this,b),l()}}(k[n])),g(a[n],j,b,n,h))}else!c&&l()}var h=!!a.test,i=a.load||a.both,j=a.callback||f,k=j,l=a.complete||f,m,n;c(h?a.yep:a.nope,!!i),i&&c(i)}var i,j,l=this.yepnope.loader;if(e(a))g(a,0,l,0);else if(w(a))for(i=0;i (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0];var j=d.createElement(s);var dl=l!='dataLayer'?'&l='+l:'';j.src='//www.googletagmanager.com/gtm.js?id='+i+dl;j.type='text/javascript';j.async=true;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-M677548'); Skip to main content Home About Submit ALERTS / RSS Search for this keyword Advanced Search New Results Quantitative trait loci mapping of gene expression and chromatin accessibility in primary fibroblast reveals shared allelic effects between Latin American and European ancestries View ORCID Profile Toni Boltz , View ORCID Profile Merel Bot , View ORCID Profile Sandra Lapinska , View ORCID Profile Tommer Schwarz , View ORCID Profile Kangcheng Hou , View ORCID Profile Kristina M. Garske , View ORCID Profile Malika K. Freund , Carrie E. Bearden , View ORCID Profile Gabriel Macaya , View ORCID Profile Carlos Lopez-Jaramillo , View ORCID Profile Nelson Freimer , Marco P. Boks , View ORCID Profile Rene S. Kahn , View ORCID Profile Bogdan Pasaniuc , View ORCID Profile Roel A. Ophoff doi: https://doi.org/10.1101/2025.06.09.658613 Toni Boltz 1 Department of Human Genetics, David Geffen School of Medicine , UCLA Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Toni Boltz For correspondence: tboltz{at}broadinstitute.org Merel Bot 2 Center for Neurobehavioral Genetics, Semel Institute for Neuroscience and Human Behavior , UCLA Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Merel Bot Sandra Lapinska 3 Graduate Group in Computational Biology, University of Pennsylvania Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Sandra Lapinska Tommer Schwarz 4 Department of Bioinformatics, David Geffen School of Medicine , UCLA Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Tommer Schwarz Kangcheng Hou 4 Department of Bioinformatics, David Geffen School of Medicine , UCLA Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Kangcheng Hou Kristina M. Garske 1 Department of Human Genetics, David Geffen School of Medicine , UCLA Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Kristina M. Garske Malika K. Freund 1 Department of Human Genetics, David Geffen School of Medicine , UCLA Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Malika K. Freund Carrie E. Bearden 2 Center for Neurobehavioral Genetics, Semel Institute for Neuroscience and Human Behavior , UCLA 5 Department of Psychology , UCLA Find this author on Google Scholar Find this author on PubMed Search for this author on this site Gabriel Macaya 6 Center for Research in Cellular and Molecular Biology, Universidad de Costa Rica , San José, Costa Rica Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Gabriel Macaya Carlos Lopez-Jaramillo 7 Department of Psychiatry, University of Antioquia , Medellín, Antioquia, Colombia Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Carlos Lopez-Jaramillo Nelson Freimer 2 Center for Neurobehavioral Genetics, Semel Institute for Neuroscience and Human Behavior , UCLA Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Nelson Freimer Marco P. Boks 8 Department of Psychiatry , Amsterdam UMC, Amsterdam, The Netherlands 9 Dimence Institute for Specialized Mental Health Care , Dimence Group, Deventer, The Netherlands 10 Department of Psychiatry, Brain Center University Medical Center Utrecht, University Utrecht , Utrecht, the Netherlands Find this author on Google Scholar Find this author on PubMed Search for this author on this site Rene S. Kahn 11 Department of Psychiatry, Icahn School of Medicine at Mount Sinai , New York, NY, USA Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Rene S. Kahn Bogdan Pasaniuc 12 Department of Genetics, University of Pennsylvania Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Bogdan Pasaniuc Roel A. Ophoff 1 Department of Human Genetics, David Geffen School of Medicine , UCLA 2 Center for Neurobehavioral Genetics, Semel Institute for Neuroscience and Human Behavior , UCLA Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Roel A. Ophoff Abstract Full Text Info/History Metrics Supplementary material Preview PDF ABSTRACT Quantitative Trait Locus (QTL) analysis of molecular data has identified genetic variants associated with traits such as gene expression, and colocalization of these functional QTL with GWAS risk loci has offered insights into the genetic basis of human disease. We employed gene expression (RNA-seq) and chromatin accessibility (ATAC-seq) obtained from human primary fibroblasts to investigate quantitative trait loci (QTLs) in cohorts ascertained for bipolar disorder of European (n=150) and Latin American (n=96) ancestries. Leveraging data from three countries of origin (The Netherlands, Colombia, Costa Rica) within our cohort, we characterized differences among individuals at the SNP, gene, and accessible-chromatin levels to compute ancestry-specific expression (e)QTLs and chromatin-accessibility (ca)QTLs. Across ancestries, we observed R² ≥ 0.93 for eQTL effect sizes and R² ≥ 0.95 for caQTLs, indicating a high degree of concordance. Integrating chromatin data with expression and genotype information enabled precise fine-mapping of eQTLs, yielding 203 high-confidence (posterior probability > 90 %) regulatory pathways. In downstream analyses, transcriptome-wide (TWAS) and chromatin-wide (CWAS) association studies with brain- and skin-related GWAS identified 36 TWAS-significant genes and 77 CWAS-significant open chromatin regions. These findings underscore the shared genetic regulatory mechanisms across European and Latin American ancestries, while demonstrating that ancestry-specific reference panels enhance the accuracy of TWAS and CWAS in diverse populations. INTRODUCTION In recent years, genome-wide association studies (GWAS) have reached sample sizes in the millions of individuals, yet nearly 80% of individuals included in these studies are of European descent. 1 Similarly, large-scale omics resources, such as GTEx, are overwhelmingly comprised of European-ancestry donors. 2 This skew limits our ability to discover genetic signals that are robust and generalizable across all populations, as variant frequencies and linkage disequilibrium (LD) patterns differ by ancestry, 3 , 4 , 5 emphasizing a need for genetic and genomic datasets from diverse populations. Incorporating data from diverse ancestries allows for the identification of both shared and population-specific genetic contributors to various traits, leading to a more comprehensive understanding of the underlying molecular mechanisms. 6 Furthermore, GWAS have successfully discovered loci associated with the risk of developing various complex diseases, including 298 independent loci currently reported for bipolar disorder. 7 However, the majority of these loci are in non-coding regions of the genome and as such, the causal mechanism between the genetic variation and disease risk often is unclear. Quantitative trait loci (QTL) studies 8 have identified non-coding SNPs that impact both gene expression and complex phenotypes, revealing mechanistic insights into disease architecture. 9 – 11 While fibroblasts are not the preferred cell type for studying psychiatric disease susceptibility, previous QTL studies have shown that cis-genetic effects are generally shared across cell types, with the caveat that brain tissue types are more highly correlated with each other than non-brain tissue types. 12 , 13 Relatedly, previous studies have shown that cis-effects on chromatin accessibility tend to be less context-dependent than on gene expression, 14 thus we expect better power to detect associations despite the tissue type. Given the ease of accessibility of a skin sample relative to a brain tissue sample, fibroblasts provide a unique opportunity to study biological samples from cohorts of hundreds of individuals. We consider fibroblasts to be a better tissue for studying neuropsychiatric traits than blood given the closer developmental origin (fibroblasts and neurons from the ectoderm, blood from the mesoderm) and greater transcriptomic stability of fibroblasts, compared to various blood cell types that require rapid response to immune or environmental perturbations. In this study, we perform a QTL mapping study in fibroblasts of a multi-ancestry sample with multi-omics data. Specifically, we investigate human primary fibroblasts derived from skin biopsies, originating from participants from The Netherlands (n=150), Colombia (n=50), and Costa Rica (n=46), with SNP-genotypes, gene expression via RNA-seq, and chromatin accessibility via ATAC-seq 15 (Assay for Transposon-Accessible Chromatin sequencing) data measured for all individuals. We use this uniquely large and diverse functional genomics dataset to evaluate effects of quantitative traits and their association with GWAS risk loci. We found that eQTL and caQTL effect sizes in fibroblasts are highly concordant across European and Latin American ancestries, enabling fine-mapping of high-confidence regulatory pathways. Downstream TWAS and CWAS identified 36 genes and 77 open-chromatin regions associated with brain- and skin-related GWAS loci, underscoring shared genetic regulation and the value of ancestry-specific reference panels. RESULTS Population stratification apparent in SNP-genotypes but not in gene expression or chromatin accessibility Given the three countries of origin within our cohort, we initially characterized the differences amongst these individuals at the SNP, gene, and accessible-chromatin levels, which were used to compute ancestry-specific expression (e)QTLs and chromatin-accessibility (ca)QTLs. Principal component analysis (PCA) of the imputed genotypes depicted clear and significantly different ancestry-specific clusters (PC1 P ANOVA = 5.6e-16, PC2 P ANOVA = 5.7e-14 after correcting for batch year) ( Figure 1A ) (Table S1), as expected. 16 See Figure S1 for the PCA on genotypes overlaid with 1000 Genomes reference populations. However, principal component analysis of the gene expression and peak matrices revealed stronger correlations with batch year than with ancestry (Table S1) ( Figures 1B and 1C ). This suggests that while population stratification is clearly detectable in the SNP-genotype data, it is not as impacted at the transcriptomic or accessible chromatin levels, which are more strongly impacted by batch-related effects. Figure S2 provides the PCA plot for the expression data after correcting for year, ancestry, and other technical factors as described in the Methods ; similarly, Figure S3 provides the plot of the chromatin data after correction. The residual values after correcting for all technical and batch related covariates for gene expression and chromatin peaks are used in the downstream analyses presented here. Download figure Open in new tab Download figure Open in new tab Figure 1. Principal component analysis of genotype, expression, and chromatin datasets. (A) PCA on genotypes, each dot represents an individual. The left plot is colored by country of origin and the right plot is colored by sequencing batch year. PC1 explains 4.9% of the variance while PC2 explains 2.8%. (B) PCA on gene expression. PC1 explains 10.8% of the variance while PC2 explains 6.8%. Regarding association to ancestry group, we found that PC1 P ANOVA = 0.0027 and PC2 P ANOVA = 0.113 after correcting for batch year. (C) PCA on chromatin peaks. PC1 explains 72.7% of the variance while PC2 explains 3.2%. Regarding association to ancestry group, we found that PC1 P ANOVA = 0.45 and PC2 P ANOVA = 0.6 after correcting for batch year. Note that ATAC-seq read depth was the strongest contributing factor to the large variance in PC1. eQTL are concordant between ancestries We first identified eQTLs within each ancestry group. Within The Netherlands cohort (N=150) we identified 3,133 eGenes at an FDR of 5% (Table S2), versus 1,394 eGenes in the Costa Rica cohort (N=46) (Table S3) and 1,492 eGenes in the Colombia cohort (N=50) (Table S4). Subsetting to matching SNP-gene pairs, we found an R2 of 0.93 between the Netherlands and Costa Rica eQTLs effects, an R2 of 0.96 between The Netherlands and Colombia eQTLs effects, and an R2 of 0.95 between the Colombia and Costa Rica eQTLs effects ( Figure 2A ). Combining genotype and expression data from the ancestry groups into one cohort (N=246) allowed us to perform a meta-analysis with greater power to detect associations, and minimal population stratification (lambdaGC = 1.08). This resulted in the identification of 5,258 eGenes at 5% FDR (Table S5). Download figure Open in new tab Figure 2. Expression and chromatin-accessibility QTLs are largely concordant between European and Latin American populations. (A) Each point represents a significant SNP-gene pair (note that many genes / peaks are associated with multiple SNPs, thus may be represented in multiple points). (B) Each point represents a significant SNP-peak pair. caQTL are concordant between ancestries For chromatin-accessibility QTLs (caQTLs), we identified 2,001 peaks with FDR-significant association to genotypes in The Netherlands cohort (Table S6), 1,292 peaks in the Costa Rica cohort (Table S7), and 1,625 in the Colombia cohort (Table S8). Subsetting to matching SNP-peak pairs, we find an R2 of 0.95 between the Netherlands and Costa Rica caQTLs effects, an R2 of 0.96 between the Netherlands and Colombia caQTLs effects, and an R2 of 0.96 between the Colombia and Costa Rica caQTLs effects ( Figure 2B ). Similar to the meta-analysis eQTL mapping, combining the genotype and chromatin data from the ancestry groups into one cohort (N=246) gave better power to detect associations (lambdaGC = 1.00), resulting in 3,557 ePeaks at an FDR threshold of 5% (Table S9). Very few opposite-effect eQTLs or caQTLs were found amongst these pairwise comparisons, though these few instances are likely due to differences in LD blocks and allele frequencies between these populations. Consistency in overall direction of the effects and significance of these findings remain after pruning for SNPs in LD, both for eQTLs (Figure S4) and caQTLs (Figure S5). Causal effect correlations using local ancestry To determine the degree of concordance between causal variant effect sizes across haplotype blocks, we used admix-kit 17 to compute the genetic correlation across local ancestries, r admix , for both caQTLs and eQTLs. Prior to estimation of r admix , we utilized ADMIXTURE 18 to determine ancestries of interest for our admixed individuals, which showed that the average admixture proportions of the Costa Rican and Colombian individuals were approximately 62% European, 33% Latin American, 5% West African, and 1% East Asian ( Figure 3 ). Given that European and Latin American haplotypes made up the vast majority (>95%) of the ancestry admixture for these individuals, a two-way admixture was performed on 97 admixed individuals selected based on joint PCA with populations from the 1000 Genomes reference panel. Download figure Open in new tab Figure 3. Proportions of reference populations via ADMIXTURE. Population reference panels from 1000 Genomes include PEL (Peruvian from Lima, Peru; AMR superpopulation), CEU (Utah residents from Northern and Western Europe; EUR superpopulation), YRI (Yoruba in Ibadan, Nigeria; AFR superpopulation), and CHB (Han Chinese in Beijing, China; EAS superpopulation). Then, given our assumption that cis-SNPs are more likely to affect expression than trans-SNPs, we obtained r admix estimates for 39,850 caQTLs and 11,523 eQTLs. After exclusion of genes or peaks with low standardized heritability (<2.0) and low confidence interval (CI) widths (<0.5), results for 1,000 eQTLs (Table S10) and 987 caQTLs (Table S11) were meta-analyzed to obtain r admix = 0.983 (C.I: [0.97, 0.996], p-value = 0.019) for caQTLs and r admix = 0.956 (C.I: [0.943, 0.968], p-value = 1e-12) for eQTLs. This suggests that the causal allelic effect sizes are the same across ancestries for these QTL. Fine-mapping of causal pathways In order to identify potential causal paths starting from SNPs impacting the accessibility of a chromatin region which in turn leads to changes in expression of a proximal gene, we used the pathfinder 19 framework. Briefly, pathfinder uses a hierarchical statistical framework to fine-map SNPs with chromatin marks and chromatin marks with gene expression in order to predict causal paths from SNP to mark to gene expression. Given the paired-sample design of this analysis, we decided to continue only with the pooled analysis, rather than limit power by subsetting each ancestry group. We defined 100kb regions centered around the transcriptional start sites (TSS) of the 5,258 eGenes identified in the meta-analysis eQTL analysis, though we filtered these down to regions with prior evidence of gene-chromatin and chromatin-SNP associations (see Methods ), resulting in a total of 934 gene regions. This low retention of regions is consistent with the empirical data analysis performed in the original pathfinder study (17.7% here and 8.9% in Roytman et. al 19 ). Of these 934 tested gene regions, we found 203 genes with a posterior probability (PP) over 90% contained within the top ten paths in the region, and of these, 28 genes with PP>=90% in a single SNP-peak pair, suggesting high confidence that the expression of these genes is causally impacted by the fine-mapped SNP and chromatin mark (Table 1 for top paths, Table S12 for all paths). We also investigated the spatial relationships between the SNPs, peaks, and genes in the top paths compared to a down-sampled set of random paths. Specifically, all paths with PP>=90% contained within the top 10 SNP/peak pairs for a gene were included as top paths, resulting in 1,893 paths. After randomly selecting 1,893 non-top paths, we plotted and compared the distributions of sizes between SNP-peak, SNP-gene, and peak-gene pairs (Figure S6). Two sample t-tests suggested no significant difference in SNP-peak distances between top paths and randomly generated paths (P=0.72), though significant differences were observed for SNP-gene (p=1.8e-15) and peak-gene (p<2.2e-16) distances. Top paths had significantly shorter distances between SNPs and genes compared to random paths. However, peak-gene distances were significantly longer in top paths compared to random paths, suggesting that these regions of open chromatin are enhancers to more distant genes. Transcriptome-wide associations to brain and skin-related traits We performed a transcriptome-wide association study (TWAS) via the FUSION 20 framework. Given the differences in ancestral background within the cohort, we split the analysis into European (Netherlands) and Latin American (Colombia and Costa Rica together) subsets. The ancestry-specific analysis resulted in 412 heritable eGenes from the European analysis, and 508 heritable eGenes from the Latin American analysis, with 67 genes overlapping between the two. Fisher’s exact test indicates this is a significant overlap (P = 6.2e-13), suggesting high confidence in the reproducibility across ancestries for these eGenes. For association to GWAS risk loci, we tested summary statistics for bipolar disorder 7 (multi-ancestry, though mostly European) and SCZ (both European 21 and Latin American 22 ancestry) as well as fibroblast-related traits including the UK BioBank 23 dermatological disease traits eczema and psoriasis. We opted to include skin-related traits given that the functional data is collected from primary fibroblasts, and brain-related traits given the inclusion of individuals diagnosed with bipolar disorder. The association testing was performed separately per ancestry, then resulting Z scores were meta-analyzed via inverse variance weighting (IVW). Pooling data across ancestries has been shown to lead to false positives, 24 though we chose to include this strategy to compare results across the different approaches. For bipolar disorder, we identified two significant genes with P IVW <= 1e-4 (Bonferroni-corrected for 500 heritable genes), including LRP5L (P IVW = 2.4e-7) and RPL27A (P IVW = 1e-5). A total of eight genes were significantly associated with the European-ancestry SCZ GWAS, including one gene shared with the BD-TWAS, LRP5L (P IVW = 8.1e-8). Other genes identified using the European-ancestry SCZ GWAS included GABARAP (P IVW = 9.9e-8), LRRC37A4P (P IVW = 4.4e-7), PLD2 (P IVW = 1.5e-6), RNF169 (P IVW = 2.1e-6), ILF2 (P IVW = 2.5e-6), TCHP (P IVW = 2.7e-5), and PMM1 (P IVW = 8.8e-5). LRRC37A4P was listed in the PGC schizophrenia study as an SMR-significant prioritized gene, based on PsychENCODE gene expression data. No genes were significantly associated with the Latin American-ancestry SCZ GWAS after Bonferoni correction (likely due to the small sample size of the GWAS). However, seven genes were nominally significant with the Latin American-ancestry GWAS, including NCAM1 which was also nominally associated (0.05 > P > 1e-4) with the European ancestry SCZ and BD GWASes. For the skin-related traits, we found three genes significantly associated with eczema, including FBL (P IVW = 7.1e-5), PMM1 (P IVW = 2.1e-5) and ILF2 (P IVW = 9.0e-7), with the latter two interestingly also associated with schizophrenia in our TWAS analysis above. We did not find any Bonferroni-significant genes associated with psoriasis. See Figure 4A for TWAS results with all tested traits after IVW meta analysis. Download figure Open in new tab Download figure Open in new tab Figure 4. Transcriptome-wide (A) and chromatin-wide (B) association analyses with brain and skin-related traits. Values in the cells refer to −log10(p-value), with darker shades of red corresponding to lower p-values. Gene names are annotated for the rows of the TWAS heatmap. Rows of the CWAS heatmap are annotated with the most proximal gene and the open chromatin region boundaries. When comparing the ancestry-specific results or the IVW-meta analysis results with the pooled-analysis TWAS, we found almost no significant eGenes in common for any traits, providing further evidence that pooled-ancestry analysis may be unreliable in the context of TWAS. Only NCAM1 was significantly associated with European-ancestry SCZ both in the pooled analysis and IVW-meta analysis (though only nominally associated). See Table S13 and S14 for meta analysis and per-ancestry TWAS results, respectively, including SNP-based heritability for each gene. Chromatin-wide associations to brain and skin-related traits We performed a chromatin-wide association study (CWAS, 10 also called cis-trome wide) by again leveraging the FUSION-TWAS framework to associate regulatory elements with GWAS summary statistics. For association to GWAS risk loci, we used summary statistics for the same brain-related traits and dermatological traits as the TWAS analysis. Also paralleling the TWAS analysis, the association testing was performed separately per ancestry then meta-analyzed via inverse variance weighting (IVW). We identified four P IVW- significant open chromatin regions with association to bipolar disorder (corresponding to four different proximal genes, see Methods ). Two of these regions were proximal to brain-expressed genes ( SYT14 and SGIP1 ). We identified 16 significantly associated open chromatin regions for European-ancestry schizophrenia (across 11 proximal genes); and no associated open chromatin regions for Latin American-ancestry schizophrenia. Of the genes that we identified to be proximal to these open chromatin regions, only KANSL1 (corresponding to open chromatin region chr17:44182964-44183252bp) has shown prior evidence of association to schizophrenia, specifically using fetal brain gene expression from PsychENCODE. For the skin-related traits, we found 12 open chromatin regions (across 12 proximal genes) significantly associated with psoriasis, and 24 chromatin regions significantly associated with eczema (across 23 proximal genes). See Figure 4B for CWAS results after IVW meta analysis. See Table S15 for meta-analysis results and S16 for per-ancestry results, including peak heritability. DISCUSSION In this study, we performed QTL analysis on gene expression and chromatin accessibility data from human primary fibroblasts of a multi-ancestry bipolar disorder sample set. QTL effect sizes were found to be concordant between the European and admixed American populations, suggesting that the genetic variants that affect either gene expression or chromatin accessibility tend to have similar effects across these populations. However, while the effect sizes may be consistent, the underlying genetic architecture and linkage disequilibrium (LD) patterns can vary significantly between ancestries, 25 contributing to the clustering of ancestral groups in principal component analysis of genetic data. For more reliable integration of molecular data into GWAS, it is necessary to match the genetic ancestral background to functional omics datasets and LD reference panels. This approach not only improves the accuracy of genomic studies 25 , 1 but also fosters inclusivity and furthers genomic research efforts on a global scale. Multi-omics datasets have advanced the field of molecular biology by providing a comprehensive and integrated view of biological systems at various levels. By collecting genomics, transcriptomics, and chromatin data across hundreds of individuals, we and others have shown that such datasets can unravel molecular mechanisms underlying biological processes. Such comprehensive analyses facilitate in deciphering complex biological patterns and predicting interactions at the molecular level. Regarding our TWAS and CWAS analyses, we find more than twice the number of chromatin peaks (87) than genes (36) associated with the tested traits. This is consistent with the prostate cancer findings in the original CWAS method paper, 10 providing further evidence to the idea that gene expression tends to be more context-dependent than chromatin accessibility 14 and that GWAS loci tend to be depleted of eQTLs. 11 As an illustrative example of the merits of combining multi-omics data from a single cohort, we were able to finemap a possible pathway for the MICA gene (major histocompatibility complex (MHC) class I polypeptide-related sequence A). MICA has been previously found to be associated with schizophrenia, 26 , 27 and integrating the expression data for this gene region with chromatin data suggested a causal pathway with posterior probability >0.98 from rs2442724 to a locus of open chromatin in the MHC region (chr6:31,367,110-31,369,941), upstream of the MICA gene (chr6:31,368,488-31,383,092). While the MHC region is known to be difficult to parse given extensive LD, 28 the large cohort of paired transcriptome/cistrome data allowed for fine-mapping of the regulation of this particular gene. Several MHC genes have been previously reported as associated with psychiatric disorders, 29 , 30 – 32 implicating this region as a probable risk locus. Another gene, Long Intergenic Non-Protein Coding RNA 933 (ID: ENSG00000259728, genomic location chr15:85,114,155-85,121,355), for which we identified a highly probable (PP=0.99) pathway has been previously associated with BD 33 and SCZ 21 via the PsychENCODE TWAS. 34 The causal path included SNP rs12900391 with a chromatin peak at chr15:84,542,517-84,544,020. While the exact function of this gene is not well understood, the genetic associations with BD found by the latest wave of the Psychiatric Genomics Consortium have suggested the involvement of pathways regulating insulin secretion, calcium channel activity, and signaling of endocannabinoids and glutamate receptors. 33 We also found a probable pathway for the gene Gamma-Aminobutyric Acid (GABA) Receptor Subunit Rho-2 ( GABRR2) . GABA is the major inhibitory neurotransmitter in all mammalian brains, and while direct association between BD or SCZ and this gene has not yet been detected, the locus of this gene on chromosome 6q has had associations to psychiatric illness. 35 – 37 Similarly, we identified a probable pathway for the GATAD2A gene, previously found to be associated with SCZ in the latest GWAS. 21 We identified probable pathways for other interesting brain-relevant genes including CRCP , which encodes a membrane protein that functions as part of a receptor complex for a small neuropeptide that increases intracellular cAMP levels in the brain. 38 While we identify novel molecular mechanisms for genes potentially relevant to disease biology, there are several limitations to our study. Although fibroblasts can offer valuable insights into certain biological processes due to shared cis-genetic effects, their relevance to complex brain-specific pathways remains limited. However, fibroblasts have previously been used in studying circadian rhythms, a phenotype which is known to be dysregulated in bipolar disorder, suggesting that relevant molecular mechanisms are at least partially preserved. 39 , 40 Secondly, this investigation focused solely on cis-QTLs, while trans-QTLS, though more difficult to ascertain, could potentially unveil additional regulatory elements involved in the brain and skin-related traits studied here. Furthermore, while our study is an important step forward in the inclusion of Latin American individuals in genomic studies, the lack of samples from individuals of African and Asian countries and other diverse ancestries represents a limitation, as genetic variants that are common to these populations may be missing or very rare in the European and Latin American samples included here. Lastly, the scarcity of Latin American-based GWAS datasets on which to perform ancestry-matched TWAS underscores the urgent need for more extensive efforts to incorporate diverse ancestral backgrounds in genetic studies. Utilizing GWAS and functional data from diverse ancestries is crucial for understanding the genetic basis of complex traits in a globally inclusive manner. METHODS Sequencing data collection Skin biopsies were collected in 2010-2012 from individuals of Colombian ancestry and Costa Rican ancestry, and collected in 2013-2014 from individuals of Dutch ancestry to generate human primary fibroblasts. Primary fibroblast cultures were established following standard procedures 41 and stored as frozen aliquots in liquid nitrogen. Fibroblasts were thawed out in batches of 8 lines at a time and grown to confluence in T75 culture flasks in standard culture media (DMEM containing 10% fetal bovine serum (FBS) and 1x Penicillin-Streptomycin). Upon reaching confluence, cells were passaged to 12 well plates at a density of 1×10 5 for ATAC and 2×10 5 for RNA. The next day cells were collected for further processing. This cohort includes BD patients and unrelated healthy controls. DNA and RNA extractions from these cells were performed simultaneously from the same batch of cells in order to minimize technical artifacts that may confound the mediation analysis. Relatedly, nine of the Netherlands samples were sequenced for RNA-seq and ATAC-seq in both batches and correlation of the gene expression and peak intensities was high, particularly for gene expression (Table S17). For duplicated sample pairs with high correlation (R2>0.85) for both gene expression and chromatin peak intensity, we randomly selected one in the pair of IDs to include in downstream analyses. For the two pairs with low correlation in chromatin peak intensities, quality control revealed that the second batch had significantly lower read depth for these individuals, thus we included the samples originating from the first batch in these instances. RNA-seq data generation and processing RNA was extracted from fibroblasts in order to assess the levels of gene expression across the genome. Cells were lysed using 350uL RLT lysis buffer from the Qiagen RNeasy mini kit. Lysed cells were then scraped off the plate, transferred to a Qiaschredder (Qiagen 79656) and centrifuged for 2 min at max speed to further homogenize. Cell lysates were kept in −80 until extraction. RNA from cell lysates was extracted using the Qiagen RNeasy mini kit (Qiagen 74106). Cell lysates were extracted in a randomized order to prevent batch effects in downstream analysis. In order to collect total RNA including small RNAs, the standard extraction protocol (Purification of Total RNA from Animal Cells using Spin Technology) was adjusted by making the following changes: - adding 1.5 volumes of 100% ethanol, instead of 70%, after the lysis step (step 4 in handbook protocol) - adding 700 mL of buffer RWT (Qiagen 1067933) instead of the provided RW1 (step 6 in handbook protocol) TruSeq Stranded polyA selected library preps were generated and samples were sequenced on the Illumina HiSeq 4000 sequencer with 75-base paired end reads, at an average of 50 million mapped reads per sample. The resulting FASTQ files were pseudo-aligned to hg19 using kallisto, 42 resulting in a matrix of transcripts per million (TPMs) which were aggregated to the gene level. The expression matrix was filtered for protein-coding genes and outliers based on technical variation (n genes = 18,886 remain after filters). Principal component analysis was initially performed on the log-transformed counts matrix and resulting PCs were assessed for correlation with batch year and ancestry. For downstream analyses, we corrected the expression matrix by including covariates for sex, first two genotype PCs, and the year of the sequencing batch (for Netherlands individuals). Previous studies have shown that there are likely unmeasured or “hidden” factors that reduce the power to detect associations in next-generation sequencing data, therefore we also performed PEER 43 (Probabilistic Estimation of Expression Residuals) factor analysis to find such hidden determinants of variation. PEER factors were computed separately per ancestry group, including 5 factors for the Colombian and Costa Rican groups, and 15 for the Netherlands group, given the difference in sample sizes. The resulting residuals matrix that remained after accounting for all factors was then used as input for downstream analyses. ATAC-seq data generation and processing ATAC-seq libraries were generated as previously described. 44 Samples were sequenced on the Illumina HiSeq 4000 sequencer with 75-base paired end reads, at an average of 39 million mapped reads per sample. Trimmed reads were aligned to the hg19 reference genome using bowtie2 45 , and filtering steps were taken to remove unmapped reads, non-primary alignment, and low-quality reads, as recommended by the ENCODE standards for ATAC-seq data analysis. 46 Reads were then input into MACS2 47 in order to call peaks that are significantly enriched against the local background using a false-discovery rate (FDR) correction threshold of 0.05 and modeled by the Poisson distribution. From this initial set of peaks, blacklisted regions 48 as identified by ENCODE were removed in order to exclude regions that have anomalous or unusually high signals due to repetitive or unstructured sequences. Then, the remaining peaks across all samples were combined to form the consensus regions, using the bedtools intersect function to stitch together any regions within 147bp into one larger peak region. This resulted in over 418,000 consensus peaks, which were limited to only those peaks that are called in at least 30% of individuals, reducing the number of peaks to 77,957. These consensus regions, in conjunction with the filtered reads (bam files) per individual, were used in the R package featureCounts to determine the number of reads each individual had within each peak, with higher counts of reads per peak indicating greater accessibility of that region of the genome. This resulted in an N x P peak counts matrix where N = number of individuals and P = number of peak regions. This matrix was log-transformed to account for skewness and ensure normalization. Principal component analysis (PCA) was then performed on the log-transformed counts matrix to assess for batch-specific and ancestry-specific correlations with PCs. PCs were correlated via Spearman’s rank correlation against other technical factors in order to determine drivers of variation within the data that are unrelated to underlying biology. Using a Bonferroni significance level of P < 2e-05, we found that sex, read depth, fraction of mitochondrial DNA, TSS enrichment score, median fragment size, and fraction of reads in peaks were correlated with the first two ATAC-seq PCs. The sequencing batch year was also included in order to account for batch effects. Using the log-transformed counts matrix as the input measures and the independent technical factors identified from the PC correlations plus age as covariates, we used PEER to find hidden confounders. PEER factors were computed separately per ancestry group, including 5 factors for the Colombian and Costa Rican groups, and 15 for the Netherlands group, given the difference in sample sizes. The resulting residuals matrix that remained after accounting for all factors was then used as input for the QTL analysis. Genotyping and imputation Samples were genotyped and imputed separately in two batches, the first consisting of 129 individuals of European ancestry genotyped via the OmniExpressExome platform, with the second batch consisting of 21 individuals of Dutch ancestry, 50 individuals of Colombian ancestry, and 46 individuals of Costa Rican ancestry genotyped via the Global Screening Array. Genotypes were first filtered for Hardy-Weinberg equilibrium p value < 1.0e-6 for controls and p value 0.01. Genotypes were then imputed into the 1000 Genomes Project phase 3 49 reference panel by chromosome using RICOPILI v.1 50 separately per genotyping platform, then subsequently merged, applying an individual-missingness threshold of 10%, SNP-missingness of 5%, and MAF > 0.05 for post-merge quality control. Imputation quality was assessed by filtering variants where genotype probability > 0.8 and INFO score > 0.1 resulting in 2,747,786 autosomal SNPs that were common across both datasets. QTL mapping QTL analysis was performed with MatrixEQTL 51 using a cis-locus distance defined as +/-1Mb around the peak midpoint or gene TSS, initially done separately per ancestry group. We included the identity by state (IBS) similarity matrix of the genotypes as an error covariance term within the model. Associations that remain after an FDR threshold of 5% were retained for downstream analysis. Correlations above R2 of 90% of the resulting SNP-gene or SNP-peak effect sizes per ancestry group suggested that the groups could be combined for a gain in power, and thus the analysis was repeated with all individuals together and with an added covariate term for ancestry. Assessing ancestry-specific causal effects Ancestry specific causal effects were calculated using admix-kit 17 on 97 admixed individuals, including 45 Costa Ricans, 50 Colombians, and 2 Dutch. These admixed individuals were determined by performing a joint PCA with the imputed genotype and 1000 Genomes Project reference panel. 49 The first 4 PCs, maximum sample distance of 1.5, sample t-range of (0.05, 0.95), and super populations EUR and AMR were used via the select-admix-indiv function in admix to select individuals to include in our analysis. To determine the super populations as well as the ancestries to use for the correlation estimates, we utilized ADMIXTURE 18 to compare our cohort with four reference populations, including Europeans (CEU), West Africans (YRI), Latin Americans (PEL), and East Asians (CHB) from 1000 Genomes Project to determine admixture proportions with supervised analysis. With these admixed individuals, we inferred local ancestry using RFmix. 52 The admix-kit package was used to determine the similarity among the genome-wide causal allelic effects across specified local ancestries in admixed individuals. 17 To calculate genetic correlations for each gene or peak, we built a window-based genetic relationship matrix, GRM, using a cis-locus distance defined as +/- 1Mb around each peak or gene rather than a genome-wide GRM to focus on the contribution of cis-SNPs on expression. The window-based GRM is used to estimate log-likelihood at different r admix values to obtain the point estimate, credible interval, and p-value for each gene or peak. To obtain the genetic correlation across all genes or peaks, we meta-analyze across the genes/peaks via the meta-analyze-genet-cor() function in admix. Prior to meta-analysis, due to our low sample size, we exclude genes or peaks whose standardized heritability (hsq_est/hsq_stderr) is less than 2, confidence interval widths are below 0.5, and genes/peaks that had more than one credible interval. Fine-mapping of SNP-chromatin-gene pathways We used the pathfinder method which accounts for both SNP LD and the correlation structure between chromatin marks by using a multivariate normal distribution. This method iterates through each possible path to determine its corresponding posterior probability, thus enabling us to prioritize SNPs and chromatin peaks that mediate gene expression, which can then be prioritized for functional validation. We restricted the regions tested by taking the transcriptional start site (TSS) for each eGene and pulling out all SNPs and chromatin peaks within 50kb upstream or downstream of the TSS. These regions were then filtered via a two-stage regression analysis, wherein the gene expression values were regressed on the proximal chromatin marks, and for models with a resulting p-value less than 0.05, we regressed SNP genotypes in that region onto the residuals from the initial peak-gene regression. Any regions with at least one p-value less than 0.05/(number of SNPs in region) were retained for pathfinder analysis. All chromatin peaks within the region were correlated pairwise via the cor() function in R, and LD for all SNPs within the region was calculated through plink –r2 square. We then inputted these regions into the pathfinder.R script ( https://github.com/meganroytman/pathfinder/blob/master/pathfinder.R ). TWAS To identify eQTL associated with GWAS traits, we performed a TWAS using the FUSION 20 software ( http://gusevlab.org/projects/fusion ). First, we generated weights for all 5,258 FDR-significant eGenes using the FUSION.compute_weights.R script, restricted to loci +/- 1Mb around the lead SNPs per each gene. We used the PEER-corrected gene expression thus no additional covariates were included in the model. In generating the weights, eGenes were first filtered for those with significant SNP-heritability (p-value <= 0.05). We used the FUSION.assoc_test.R script to test for association between the gene weights and GWAS for bipolar disorder, schizophrenia (EUR and AMR ancestry), and UKBB skin-related GWAS for eczema and psoriasis. Gene-trait pairs were selected based on the best performing model after five-fold cross validation, including for Best Unbiased Linear Predictor (BLUP), elastic net (ENET), Least Absolute Shrinkage and Selection Operator (LASSO), and just using the top SNP. To account for LD structure we used an in-sample LD panel. The --coloc flag was included to perform colocalization 53 on any genes that had an association with the trait of interest with TWAS.P < 0.05. CWAS Similarly, to identify caQTL associated with GWAS traits, we performed a CWAS 10 using the FUSION software. First, we generated weights for all FDR-significant ePeaks using the FUSION.compute_weights.R script, restricted to loci +/- 1Mb around the lead SNPs per each gene. We used the PEER-corrected peak intensities thus no additional covariates were included in the model. We used the FUSION.assoc_test.R script to test for association between the peak weights and GWAS for bipolar disorder, schizophrenia (EUR and AMR ancestry), and UKBB skin-related GWAS for eczema and psoriasis. Peak-trait pairs were selected based on the best performing model after five-fold cross validation, including for Best Unbiased Linear Predictor (BLUP), elastic net (ENET), Least Absolute Shrinkage and Selection Operator (LASSO), and just using the top SNP. To account for LD structure we used an in-sample LD panel. The --coloc flag was included to perform colocalization on any peaks that had an association with the trait of interest with TWAS.P < 0.05. Proximal genes were identified using the GenomicRanges 54 package in R, employing the nearest() function. Supplemental Figures and Table Captions Figure S1: Principal component analysis of sample cohort overlaid onto 1KG. EUR = European, AFR = African, EAS = East Asian, SAS = South Asian, AMR = American, SAMPLE = cohort analyzed in this study. Figure S2: Principal component analysis of gene expression data after correction for covariates. Annotated by country of origin on the left and year of sequencing batch on the right. Figure S3: Principal component analysis of chromatin accessibility data after correction for covariates. Annotated by country of origin on the left and year of sequencing batch on the right. Figure S4: Pairwise comparison of effect sizes between ancestry-specific eQTLs. Only including LD-independent SNPs Figure S5: Pairwise comparison of effect sizes between ancestry-specific caQTLs. Only including LD-independent. Figure S6: Spatial relationships between SNPs, peaks, and genes in top vs random paths. A. SNP-peak distances; B. SNP-gene distances; and C. Peak-Gene distances. Table 1: Genes with posterior probability of paths >=90% in a single SNP-peak pair. Peaks are formatted as chr:startBP-endBP. Table S1: Variance explained and Analysis of Variance (AOV) p values for first 5 principal components of genotype, expression, and chromatin data with ancestry and sequencing year. Table S2: Top eQTLs for Netherlands cohort, including SNP, ENSG, p-value and beta effect size. Table S3: Top eQTLs for Costa Rica cohort, including SNP, ENSG, p-value and beta effect size. Table S4: Top eQTLs for Colombia cohort, including SNP, ENSG, p-value and beta effect size. Table S5: Pooled-cohort eQTL results. Table S6: Top caQTLs for Netherlands cohort, including SNP, peak boundaries, p-value and beta effect size. Table S7: Top caQTLs for Costa Rica cohort, including SNP, peak boundaries, p-value and beta effect size. Table S8: Top caQTLs for Colombia cohort, including SNP, peak boundaries, p-value and beta effect size. Table S9: Pooled-cohort caQTL results. Table S10: Admixture results for tested genes. rg_mode refers to the genetic correlation between the local ancestry haplotypes of the eQTL; conf_inf refers to the confidence interval around the rg_mode values. Table S11: Admixture results for tested peaks. rg_mode refers to the genetic correlation between the local ancestry haplotypes of the caQTL; conf_inf refers to the confidence interval around the rg_mode values. Table S12: All paths for genes having at least 90% PP within its top 10 SNP/peak pairs. Peaks are formatted as chr:startBP-endBP. Posteriors refer to posterior probability. Table S13: Meta-analysis of TWAS results for brain and skin-related traits. Beta = IVW-meta analysis P value; se = IVW-meta analysis standard error; z = IVW-meta analysis z score; p = IVW-meta analysis p value; SumStats = cohort/trait summary statistics used. Table S14: TWAS-FUSION results for ancestry-specific and pooled analysis. Abbreviations: CHR; chromosome, P0; gene start, P1; gene end, HSQ; heritability of the gene, BEST.GWAS.ID; rsID of the most significant SNP in locus, BEST.GWAS.Z; Z-score of the most significant GWAS SNP in locus, EQTL.ID; rsID of the best eQTL in the locus, EQTL.R2; cross-validation R2 of the best eQTL in the locus, EQTL.Z; Z-score of the best eQTL in the locus, EQTL.GWAS.Z; GWAS Z-score for this eQTL, NSNP; number of SNPs in the locus, MODELCV.R2; cross-validation R2 of the best performing model, MODELCV.PV; cross-validation P-value of the best performing model, SNP; single nucleotide polymorphism, FDR; false-discovery rate, eQTL; expression quantitative trait locus. PP; posterior probability (0: no association, 1: eQTL association only, 2: trait association only, 3: independent eQTL/trait association, 4: colocalized association); SumStats = cohort/trait summary statistics used. Table S15: Meta-analysis of CWAS results for brain and skin-related traits. Beta = IVW-meta analysis P value; se = IVW-meta analysis standard error; z = IVW-meta analysis z score; p = IVW-meta analysis p value; SumStats = cohort/trait summary statistics used; Closest.Gene = gene symbol for nearest gene to chromatin region. Table S16: CWAS-FUSION results for ancestry-specific and pooled analysis. Abbreviations: CHR; chromosome, P0; chromatin start, P1; chromatin end, HSQ; heritability of the peak, BEST.GWAS.ID; rsID of the most significant SNP in locus, BEST.GWAS.Z; Z-score of the most significant GWAS SNP in locus, EQTL.ID; rsID of the best caQTL in the locus, EQTL.R2; cross-validation R2 of the best caQTL in the locus, EQTL.Z; Z-score of the best caQTL in the locus, EQTL.GWAS.Z; GWAS Z-score for this caQTL, NSNP; number of SNPs in the locus, MODELCV.R2; cross-validation R2 of the best performing model, MODELCV.PV; cross-validation P-value of the best performing model, SNP; single nucleotide polymorphism, FDR; false-discovery rate, caQTL; chromatin accessibility quantitative trait locus. PP; posterior probability (0: no association, 1: caQTL association only, 2: trait association only, 3: independent caQTL/trait association, 4: colocalized association); SumStats = cohort/trait summary statistics used; Closest.Gene = gene symbol for nearest gene to chromatin region. Table S17: Duplicated Netherlands sample correlation for gene expression and chromatin accessibility data. RNA R2 = correlation between gene expression TPMs of the two samples; ATAC R2 = correlation between chromatin accessible peaks of the two samples. REFERENCES 1. ↵ Martin , A. R. et al. Current clinical use of polygenic scores will risk exacerbating health disparities . Nat. Genet . 51 , 584 ( 2019 ). OpenUrl CrossRef PubMed 2. ↵ Yang , G. , Mishra , M. & Perera , M. A . Multi-Omics Studies in Historically Excluded Populations: The Road to Equity . Clin. Pharmacol. Ther . 113 , 541 – 556 ( 2023 ). OpenUrl CrossRef PubMed 3. ↵ Goddard , K. A. , Hopkins , P. J. , Hall , J. M. & Witte , J. S . Linkage disequilibrium and allele-frequency distributions for 114 single-nucleotide polymorphisms in five populations . Am. J. Hum. Genet . 66 , ( 2000 ). 4. ↵ Pritchard , J. K. & Donnelly , P . Case-control studies of association in structured or admixed populations . Theor. Popul. Biol . 60 , ( 2001 ). 5. ↵ Hellwege , J. et al. Population Stratification in Genetic Association Studies . Curr. Protoc. Hum. Genet . 95 , 1.22.1 ( 2017 ). OpenUrl CrossRef 6. ↵ Shi , H. et al. Localizing Components of Shared Transethnic Genetic Architecture of Complex Traits from GWAS Summary Data . American Journal of Human Genetics 106 , 805 ( 2020 ). OpenUrl CrossRef PubMed 7. ↵ O’Connell , K. S. et al. Genomics yields biological and phenotypic insights into bipolar disorder . Nature 1 – 12 ( 2025 ). 8. ↵ Nica , A. C. & Dermitzakis , E. T . Expression quantitative trait loci: present and future . Philos. Trans. R. Soc. Lond. B Biol. Sci . 368 , ( 2013 ). 9. ↵ Gusev , A. et al. Transcriptome-wide association study of schizophrenia and chromatin activity yields mechanistic disease insights . Nat. Genet . 50 , 538 – 548 ( 2018 ). OpenUrl CrossRef PubMed 10. ↵ Baca , S. C. et al. Genetic determinants of chromatin reveal prostate cancer risk mediated by context-dependent gene regulation . Nat. Genet . 54 , 1364 – 1375 ( 2022 ). OpenUrl CrossRef PubMed 11. ↵ Mostafavi , H. , Spence , J. P. , Naqvi , S. & Pritchard , J. K . Systematic differences in discovery of genetic effects on gene expression and complex traits . Nat. Genet . 55 , 1866 – 1875 ( 2023 ). OpenUrl CrossRef PubMed 12. ↵ Qi , T. et al. Identifying gene targets for brain-related traits using transcriptomic and methylomic data from blood . Nat. Commun . 9 , 1 – 12 ( 2018 ). OpenUrl CrossRef PubMed 13. ↵ GTEx Consortium . Genetic effects on gene expression across human tissues . Nature 550 , 204 – 213 ( 2017 ). OpenUrl CrossRef PubMed Web of Science 14. ↵ Alasoo , K. et al. Shared genetic effects on chromatin and gene expression indicate a role for enhancer priming in immune response . Nat. Genet . 50 , 424 ( 2018 ). OpenUrl CrossRef PubMed 15. ↵ Buenrostro , J. , Wu , B. , Chang , H. & Greenleaf , W . ATAC-seq: A Method for Assaying Chromatin Accessibility Genome-Wide . Curr. Protoc. Mol. Biol . 109 , 21.29.1 ( 2015 ). OpenUrl CrossRef PubMed 16. ↵ Price , A. L. et al. Principal components analysis corrects for stratification in genome-wide association studies . Nat. Genet . 38 , ( 2006 ). 17. ↵ Hou , K. et al. Causal effects on complex traits are similar for common variants across segments of different continental ancestries within admixed individuals . Nat. Genet . 55 , 549 – 558 ( 2023 ). OpenUrl CrossRef PubMed 18. ↵ Alexander , D. H. , Novembre , J. & Lange , K . Fast model-based estimation of ancestry in unrelated individuals . Genome Res . 19 , ( 2009 ). 19. ↵ Roytman , M. , Kichaev , G. , Gusev , A. & Pasaniuc , B . Methods for fine-mapping with chromatin and expression data . PLoS Genet . 14 , e1007240 ( 2018 ). OpenUrl CrossRef PubMed 20. ↵ Gusev , A. et al. Integrative approaches for large-scale transcriptome-wide association studies . Nat. Genet . 48 , ( 2016 ). 21. ↵ Trubetskoy , V. et al. Mapping genomic loci implicates genes and synaptic biology in schizophrenia . Nature 604 , ( 2022 ). 22. ↵ Bigdeli , T. B. et al. Contributions of common genetic variants to risk of schizophrenia among individuals of African and Latino ancestry . Mol. Psychiatry 25 , ( 2020 ). 23. ↵ Sudlow , C. et al. UK Biobank: An Open Access Resource for Identifying the Causes of a Wide Range of Complex Diseases of Middle and Old Age . PLoS Med . 12 , e1001779 ( 2015 ). OpenUrl CrossRef PubMed 24. ↵ Best practices for multi-ancestry, meta-analytic transcriptome-wide association studies: Lessons from the Global Biobank Meta-analysis Initiative . Cell Genomics 2 , 100180 ( 2022 ). OpenUrl CrossRef PubMed 25. ↵ Martin , A. R. et al. Human Demographic History Impacts Genetic Risk Prediction across Diverse Populations . Am. J. Hum. Genet . 100 , 635 ( 2017 ). OpenUrl CrossRef PubMed 26. ↵ Mahadevan , J. et al. Analysis of whole exome sequencing in severe mental illness hints at selection of brain development and immune related genes . Sci. Rep . 11 , 1 – 10 ( 2021 ). OpenUrl CrossRef PubMed 27. ↵ Ren , X. , Mao , A. , Tan , S. , Liu , J. & Wei , X . Analysis of the association between MICA gene polymorphisms and schizophrenia . J. Clin. Lab. Anal . 36 , ( 2022 ). 28. ↵ Matzaraki , V. , Kumar , V. , Wijmenga , C. & Zhernakova , A . The MHC locus and genetic susceptibility to autoimmune and infectious diseases . Genome Biol . 18 , 1 – 21 ( 2017 ). OpenUrl CrossRef PubMed 29. ↵ Li , J. , Yoshikawa , A. , Alliey-Rodriguez , N. & Meltzer , H. Y . Schizophrenia risk loci from xMHC region were associated with antipsychotic response in chronic schizophrenic patients with persistent positive symptom . Transl. Psychiatry 12 , 92 ( 2022 ). 30. ↵ Tamouza , R. , Krishnamoorthy , R. & Leboyer , M . Understanding the genetic contribution of the human leukocyte antigen system to common major psychiatric disorders in a world pandemic context . Brain Behav. Immun . 91 , 731 ( 2021 ). 31. Schizophrenia Working Group of the Psychiatric Genomics Consortium . Biological insights from 108 schizophrenia-associated genetic loci . Nature 511 , 421 – 427 ( 2014 ). OpenUrl CrossRef PubMed Web of Science 32. ↵ Debnath , M. , Berk , M. , Leboyer , M. & Tamouza , R . The MHC/HLA Gene Complex in Major Psychiatric Disorders: Emerging Roles and Implications . Current Behavioral Neuroscience Reports 5 , 179 – 188 ( 2018 ). OpenUrl CrossRef 33. ↵ Mullins , N. et al. Genome-wide association study of more than 40,000 bipolar disorder cases provides new insights into the underlying biology . Nat. Genet . 53 , 817 – 829 ( 2021 ). OpenUrl CrossRef PubMed 34. ↵ Gandal , M. J. et al. Transcriptome-wide isoform-level dysregulation in ASD, schizophrenia, and bipolar disorder . Science 362 , ( 2018 ). 35. ↵ Kohn , Y. & Lerer , B . Excitement and confusion on chromosome 6q: the challenges of neuropsychiatric genetics in microcosm . Mol. Psychiatry 10 , ( 2005 ). 36. Dick , D. M. et al. Genomewide linkage analyses of bipolar disorder: a new sample of 250 pedigrees from the National Institute of Mental Health Genetics Initiative . Am. J. Hum. Genet . 73 , 107 – 114 ( 2003 ). OpenUrl CrossRef PubMed Web of Science 37. ↵ Middleton , F. A. et al. Genomewide Linkage Analysis of Bipolar Disorder by Use of a High-Density Single-Nucleotide–Polymorphism (SNP) Genotyping Assay: A Comparison with Microsatellite Marker Assays and Finding of Significant Linkage to Chromosome 6q22 . Am. J. Hum. Genet . 74 , 886 ( 2004 ). OpenUrl CrossRef PubMed Web of Science 38. ↵ CRCP CGRP receptor component [Homo sapiens (human)] - Gene - NCBI. https://www.ncbi.nlm.nih.gov/gene/27297 . 39. ↵ Francia , M. et al. Fibroblasts as an in vitro model of circadian genetic and genomic studies . bioRxiv 2023.05.19.541494 ( 2023 ) doi: 10.1101/2023.05.19.541494 . OpenUrl Abstract / FREE Full Text 40. ↵ Yamazaki , S. & Takahashi , J. S . Real-time luminescence reporting of circadian gene expression in mammals . Methods Enzymol . 393 , 288 – 301 ( 2005 ). OpenUrl CrossRef PubMed Web of Science 41. ↵ Villegas , J. & McPhaul , M . Establishment and culture of human skin fibroblasts . Curr. Protoc. Mol. Biol. Chapter 28 , ( 2005 ). 42. ↵ Bray , N. L. , Pimentel , H. , Melsted , P. & Pachter , L . Near-optimal probabilistic RNA-seq quantification . Nat. Biotechnol . 34 , 525 – 527 ( 2016 ). OpenUrl CrossRef PubMed 43. ↵ Stegle , O. , Parts , L. , Piipari , M. , Winn , J. & Durbin , R . Using probabilistic estimation of expression residuals (PEER) to obtain increased power and interpretability of gene expression analyses . Nat. Protoc . 7 , 500 . 44. ↵ Buenrostro , J. D. , Giresi , P. G. , Zaba , L. C. , Chang , H. Y. & Greenleaf , W. J . Transposition of native chromatin for fast and sensitive epigenomic profiling of open chromatin, DNA-binding proteins and nucleosome position . Nature Methods 10 , 1213 – 1218 ( 2013 ). OpenUrl CrossRef PubMed 45. ↵ Langmead , B. & Salzberg , S. L . Fast gapped-read alignment with Bowtie 2 . Nat. Methods 9 , 357 . 46. ↵ Hitz , B. C. , et al. The ENCODE Uniform Analysis Pipelines . bioRxiv ( 2023 ) doi: 10.1101/2023.04.04.535623 . OpenUrl Abstract / FREE Full Text 47. ↵ Zhang , Y. et al. Model-based Analysis of ChIP-Seq (MACS) . Genome Biol . 9 , 1 – 9 ( 2008 ). OpenUrl CrossRef PubMed 48. ↵ Amemiya , H. M. , Kundaje , A. & Boyle , A. P . The ENCODE Blacklist: Identification of Problematic Regions of the Genome . Sci. Rep . 9 , 1 – 5 ( 2019 ). OpenUrl CrossRef PubMed 49. ↵ 1000 Genomes Project Consortium et al. A global reference for human genetic variation . Nature 526 , 68 – 74 ( 2015 ). OpenUrl CrossRef PubMed 50. ↵ Lam , M. et al. RICOPILI: Rapid Imputation for COnsortias PIpeLIne . Bioinformatics 36 , ( 2020 ). 51. ↵ Shabalin , A. A . Matrix eQTL: ultra fast eQTL analysis via large matrix operations . Bioinformatics 28 , 1353 – 1358 ( 2012 ). OpenUrl CrossRef PubMed Web of Science 52. ↵ Maples , B. K. , Gravel , S. , Kenny , E. E. & Bustamante , C. D . RFMix: A Discriminative Modeling Approach for Rapid and Robust Local-Ancestry Inference . Am. J. Hum. Genet . 93 , 278 ( 2013 ). OpenUrl CrossRef PubMed 53. ↵ Giambartolomei , C. et al. Bayesian test for colocalisation between pairs of genetic association studies using summary statistics . PLoS Genet . 10 , e1004383 ( 2014 ). OpenUrl CrossRef PubMed 54. ↵ Lawrence , M. et al. Software for Computing and Annotating Genomic Ranges . PLOS Computational Biology 9 , e1003118 ( 2013 ). OpenUrl CrossRef View the discussion thread. Back to top Previous Next Posted June 10, 2025. Download PDF Supplementary Material Email Thank you for your interest in spreading the word about bioRxiv. NOTE: Your email address is requested solely to identify you as the sender of this article. Your Email * Your Name * Send To * Enter multiple addresses on separate lines or separate them with commas. You are going to email the following Quantitative trait loci mapping of gene expression and chromatin accessibility in primary fibroblast reveals shared allelic effects between Latin American and European ancestries Message Subject (Your Name) has forwarded a page to you from bioRxiv Message Body (Your Name) thought you would like to see this page from the bioRxiv website. Your Personal Message CAPTCHA This question is for testing whether or not you are a human visitor and to prevent automated spam submissions. Share Quantitative trait loci mapping of gene expression and chromatin accessibility in primary fibroblast reveals shared allelic effects between Latin American and European ancestries Toni Boltz , Merel Bot , Sandra Lapinska , Tommer Schwarz , Kangcheng Hou , Kristina M. Garske , Malika K. Freund , Carrie E. Bearden , Gabriel Macaya , Carlos Lopez-Jaramillo , Nelson Freimer , Marco P. Boks , Rene S. Kahn , Bogdan Pasaniuc , Roel A. Ophoff bioRxiv 2025.06.09.658613; doi: https://doi.org/10.1101/2025.06.09.658613 Share This Article: Copy Citation Tools Quantitative trait loci mapping of gene expression and chromatin accessibility in primary fibroblast reveals shared allelic effects between Latin American and European ancestries Toni Boltz , Merel Bot , Sandra Lapinska , Tommer Schwarz , Kangcheng Hou , Kristina M. Garske , Malika K. Freund , Carrie E. Bearden , Gabriel Macaya , Carlos Lopez-Jaramillo , Nelson Freimer , Marco P. Boks , Rene S. Kahn , Bogdan Pasaniuc , Roel A. Ophoff bioRxiv 2025.06.09.658613; doi: https://doi.org/10.1101/2025.06.09.658613 Citation Manager Formats BibTeX Bookends EasyBib EndNote (tagged) EndNote 8 (xml) Medlars Mendeley Papers RefWorks Tagged Ref Manager RIS Zotero Tweet Widget Facebook Like Google Plus One Subject Area Genomics Subject Areas All Articles Animal Behavior and Cognition (7635) Biochemistry (17691) Bioengineering (13892) Bioinformatics (41936) Biophysics (21452) Cancer Biology (18588) Cell Biology (25504) Clinical Trials (138) Developmental Biology (13378) Ecology (19899) Epidemiology (2067) Evolutionary Biology (24320) Genetics (15609) Genomics (22506) Immunology (17736) Microbiology (40394) Molecular Biology (17181) Neuroscience (88605) Paleontology (666) Pathology (2832) Pharmacology and Toxicology (4824) Physiology (7641) Plant Biology (15153) Scientific Communication and Education (2045) Synthetic Biology (4294) Systems Biology (9825) Zoology (2271)