An Expanded View of Complex Traits: From Polygenic to Omnigenic

A central goal of genetics is to understand the links between genetic variation and disease. Intuitively, one might expect disease-causing variants to cluster into key pathways that drive disease etiology. But for complex traits, association signals tend to be spread across most of the genome—including near many genes without an obvious connection to disease. We propose that gene regulatory networks are sufficiently interconnected such that all genes expressed in disease-relevant cells are liable to affect the functions of core disease-related genes and that most heritability can be explained by effects on genes outside core pathways. We refer to this hypothesis as an "omnigenic" model. A central goal of genetics is to understand the links between genetic variation and disease. Intuitively, one might expect disease-causing variants to cluster into key pathways that drive disease etiology. But for complex traits, association signals tend to be spread across most of the genome—including near many genes without an obvious connection to disease. We propose that gene regulatory networks are sufficiently interconnected such that all genes expressed in disease-relevant cells are liable to affect the functions of core disease-related genes and that most heritability can be explained by effects on genes outside core pathways. We refer to this hypothesis as an "omnigenic" model. The longest-standing question in genetics is to understand how genetic variation contributes to phenotypic variation. In the early 1900s, there was fierce debate between the Mendelians—who were inspired by Mendel's work on pea genetics and focused on discrete, monogenic phenotypes—and the biometricians, who were interested in the inheritance of continuous traits such as height. The biometricians believed that Mendelian genetics could not explain the continuous distribution of variation observed for many traits in humans and other species. This debate was resolved in a seminal 1918 paper by R.A. Fisher, who showed that, if many genes affect a trait, then the random sampling of alleles at each gene produces a continuous, normally distributed phenotype in the population (Fisher, 1918Fisher R.A. The correlation between relatives on the supposition of Mendelian inheritance.Trans. R. Soc. Edinb. 1918; 52: 399-433Crossref Scopus (2537) Google Scholar). As the number of genes grows very large, the contribution of each gene becomes correspondingly smaller, leading in the limit to Fisher's famous "infinitesimal model" (Barton et al., 2016Barton N.H. Etheridge A.M. Veber A. The infinitesimal model.bioRxiv. 2016; https://doi.org/10.1101/039768Google Scholar). Despite the success of the infinitesimal model in describing inheritance patterns, especially in plant and animal breeding, it was unclear throughout the 20th century how many genes would actually be important for driving complex traits. Indeed, human geneticists expected that even complex traits would be driven by a handful of moderate-effect loci—thus giving rise to large numbers of mapping studies that were, in retrospect, greatly underpowered. For example, an elegant 1999 analysis of allele sharing in autistic siblings concluded from the lack of significant hits that there must be "a large number of loci (perhaps ≥15)." This prediction was strikingly high at the time but seems quaintly low now (Risch et al., 1999Risch N. Spiker D. Lotspeich L. Nouri N. Hinds D. Hallmayer J. Kalaydjieva L. McCague P. Dimiceli S. Pitts T. et al.A genomic screen of autism: evidence for a multilocus etiology.Am. J. Hum. Genet. 1999; 65: 493-507Abstract Full Text Full Text PDF PubMed Scopus (559) Google Scholar, Weiner et al., 2016Weiner D.J. Wigdor E.M. Ripke S. Walters R.K. Kosmicki J.A. Grove J. Samocha K.E. Goldstein J. Okbay A. Bybjerg-Gauholm J. et al.Polygenic transmission disequilibrium confirms that common and rare variation act additively to create risk for autism spectrum disorders.bioRxiv. 2016; https://doi.org/10.1101/089342Google Scholar). Since around 2006, the advent of genome-wide association studies, and more recently exome sequencing, has provided the first detailed understanding of the genetic basis of complex traits. One of the early surprises of the GWAS era was that, for typical traits, even the most important loci in the genome have small effect sizes and that, together, the significant hits only explain a modest fraction of the predicted genetic variance. This has been referred to as the mystery of the "missing heritability" (Manolio et al., 2009Manolio T.A. Collins F.S. Cox N.J. Goldstein D.B. Hindorff L.A. Hunter D.J. McCarthy M.I. Ramos E.M. Cardon L.R. Chakravarti A. et al.Finding the missing heritability of complex diseases.Nature. 2009; 461: 747-753Crossref PubMed Scopus (5883) Google Scholar). The mystery has since been largely resolved by analyses showing that common single-nucleotide polymorphisms (SNPs) with effect sizes well below genome-wide statistical significance account for most of the "missing heritability" of many traits (Yang et al., 2010Yang J. Benyamin B. McEvoy B.P. Gordon S. Henders A.K. Nyholt D.R. Madden P.A. Heath A.C. Martin N.G. Montgomery G.W. et al.Common SNPs explain a large proportion of the heritability for human height.Nat. Genet. 2010; 42: 565-569Crossref PubMed Scopus (2793) Google Scholar, Shi et al., 2016Shi H. Kichaev G. Pasaniuc B. Contrasting the genetic architecture of 30 complex traits from summary association data.Am. J. Hum. Genet. 2016; 99: 139-153Abstract Full Text Full Text PDF PubMed Scopus (183) Google Scholar). Rare variants with larger effect sizes also contribute genetic variance (Marouli et al., 2017Marouli E. Graff M. Medina-Gomez C. Lo K.S. Wood A.R. Kjaer T.R. Fine R.S. Lu Y. Schurmann C. Highland H.M. et al.EPIC-InterAct ConsortiumCHD Exome+ ConsortiumExomeBP ConsortiumT2D-Genes ConsortiumGoT2D Genes ConsortiumGlobal Lipids Genetics ConsortiumReproGen ConsortiumMAGIC InvestigatorsRare and low-frequency coding variants alter human adult height.Nature. 2017; 542: 186-190Crossref PubMed Scopus (358) Google Scholar), especially for diseases with major fitness consequences (Simons et al., 2014Simons Y.B. Turchin M.C. Pritchard J.K. Sella G. The deleterious mutation load is insensitive to recent population history.Nat. Genet. 2014; 46: 220-224Crossref PubMed Scopus (176) Google Scholar) such as autism and schizophrenia (De Rubeis et al., 2014De Rubeis S. He X. Goldberg A.P. Poultney C.S. Samocha K. Cicek A.E. Kou Y. Liu L. Fromer M. Walker S. et al.DDD StudyHomozygosity Mapping Collaborative for AutismUK10K ConsortiumSynaptic, transcriptional and chromatin genes disrupted in autism.Nature. 2014; 515: 209-215Crossref PubMed Scopus (1612) Google Scholar, Fromer et al., 2014Fromer M. Pocklington A.J. Kavanagh D.H. Williams H.J. Dwyer S. Gormley P. Georgieva L. Rees E. Palta P. Ruderfer D.M. et al.De novo mutations in schizophrenia implicate synaptic networks.Nature. 2014; 506: 179-184Crossref PubMed Scopus (1127) Google Scholar, Purcell et al., 2014Purcell S.M. Moran J.L. Fromer M. Ruderfer D. Solovieff N. Roussos P. O'Dushlaine C. Chambert K. Bergen S.E. Kähler A. et al.A polygenic burden of rare disruptive mutations in schizophrenia.Nature. 2014; 506: 185-190Crossref PubMed Scopus (1001) Google Scholar). A second surprise was that, in contrast to Mendelian diseases—which are largely caused by protein-coding changes (Botstein and Risch, 2003Botstein D. Risch N. Discovering genotypes underlying human phenotypes: past successes for mendelian disease, future approaches for complex disease.Nat. Genet. 2003; 33: 228-237Crossref PubMed Scopus (1214) Google Scholar)—complex traits are mainly driven by noncoding variants that presumably affect gene regulation (Pickrell, 2014Pickrell J.K. Joint analysis of functional genomic data and genome-wide association studies of 18 human traits.Am. J. Hum. Genet. 2014; 94: 559-573Abstract Full Text Full Text PDF PubMed Scopus (330) Google Scholar, Welter et al., 2014Welter D. MacArthur J. Morales J. Burdett T. Hall P. Junkins H. Klemm A. Flicek P. Manolio T. Hindorff L. Parkinson H. The NHGRI GWAS Catalog, a curated resource of SNP-trait associations.Nucleic Acids Res. 2014; 42: D1001-D1006Crossref PubMed Scopus (1997) Google Scholar, Li et al., 2016Li Y.I. van de Geijn B. Raj A. Knowles D.A. Petti A.A. Golan D. Gilad Y. Pritchard J.K. RNA splicing is a primary link between genetic variation and disease.Science. 2016; 352: 600-604Crossref PubMed Scopus (309) Google Scholar). Indeed, many studies have shown that significant variants are highly enriched in regions of active chromatin such as promoters and enhancers in relevant cell types. For example, risk variants for autoimmune diseases show particular enrichment in active chromatin regions of immune cells (Maurano et al., 2012Maurano M.T. Humbert R. Rynes E. Thurman R.E. Haugen E. Wang H. Reynolds A.P. Sandstrom R. Qu H. Brody J. et al.Systematic localization of common disease-associated variation in regulatory DNA.Science. 2012; 337: 1190-1195Crossref PubMed Scopus (2200) Google Scholar, Farh et al., 2015Farh K.K.-H. Marson A. Zhu J. Kleinewietfeld M. Housley W.J. Beik S. Shoresh N. Whitton H. Ryan R.J. Shishkin A.A. et al.Genetic and epigenetic fine mapping of causal autoimmune disease variants.Nature. 2015; 518: 337-343Crossref PubMed Scopus (1153) Google Scholar, Kundaje et al., 2015Kundaje A. Meuleman W. Ernst J. Bilenky M. Yen A. Heravi-Moussavi A. Kheradpour P. Zhang Z. Wang J. Ziller M.J. et al.Roadmap Epigenomics ConsortiumIntegrative analysis of 111 reference human epigenomes.Nature. 2015; 518: 317-330Crossref PubMed Scopus (3578) Google Scholar). These observations are generally interpreted in a paradigm in which complex disease is driven by an accumulation of weak effects on the key genes and regulatory pathways that drive disease risk (Furlong, 2013Furlong L.I. Human diseases through the lens of network biology.Trends Genet. 2013; 29: 150-159Abstract Full Text Full Text PDF PubMed Scopus (138) Google Scholar, Chakravarti and Turner, 2016Chakravarti A. Turner T.N. Revealing rate-limiting steps in complex disease biology: The crucial importance of studying rare, extreme-phenotype families.BioEssays. 2016; 38: 578-586Crossref PubMed Scopus (33) Google Scholar). This model has motivated many studies that aim to dissect the functional impacts of individual disease-associated variants (Smemo et al., 2014Smemo S. Tena J.J. Kim K.-H. Gamazon E.R. Sakabe N.J. Gómez-Marín C. Aneas I. Credidio F.L. Sobreira D.R. Wasserman N.F. et al.Obesity-associated variants within FTO form long-range functional connections with IRX3.Nature. 2014; 507: 371-375Crossref PubMed Scopus (807) Google Scholar, Sekar et al., 2016Sekar A. Bialas A.R. de Rivera H. Davis A. Hammond T.R. Kamitaki N. Tooley K. Presumey J. Baum M. Van Doren V. et al.Schizophrenia Working Group of the Psychiatric Genomics ConsortiumSchizophrenia risk from complex variation of complement component 4.Nature. 2016; 530: 177-183Crossref PubMed Scopus (1377) Google Scholar) or to aggregate hits to identify key disease pathways and processes (Califano et al., 2012Califano A. Butte A.J. Friend S. Ideker T. Schadt E. Leveraging models of cell regulation and GWAS data in integrative network-based association studies.Nat. Genet. 2012; 44: 841-847Crossref PubMed Scopus (200) Google Scholar, Jostins et al., 2012Jostins L. Ripke S. Weersma R.K. Duerr R.H. McGovern D.P. Hui K.Y. Lee J.C. Schumm L.P. Sharma Y. Anderson C.A. et al.International IBD Genetics Consortium (IIBDGC)Host-microbe interactions have shaped the genetic architecture of inflammatory bowel disease.Nature. 2012; 491: 119-124Crossref PubMed Scopus (3230) Google Scholar, Wood et al., 2014Wood A.R. Esko T. Yang J. Vedantam S. Pers T.H. Gustafsson S. Chu A.Y. Estrada K. Luan J. Kutalik Z. et al.Electronic Medical Records and Genomics (eMEMERGEGE) ConsortiumMIGen ConsortiumPAGEGE ConsortiumLifeLines Cohort StudyDefining the role of common variation in the genomic and biological architecture of adult human height.Nat. Genet. 2014; 46: 1173-1186Crossref PubMed Scopus (1198) Google Scholar, Krumm et al., 2015Krumm N. Turner T.N. Baker C. Vives L. Mohajeri K. Witherspoon K. Raja A. Coe B.P. Stessman H.A. He Z.-X. et al.Excess of rare, inherited truncating mutations in autism.Nat. Genet. 2015; 47: 582-588Crossref PubMed Scopus (363) Google Scholar). For several diseases, the leading hits have indeed helped to highlight specific molecular processes—for example, uncovering the role of autophagy in Crohn's disease (Jostins et al., 2012Jostins L. Ripke S. Weersma R.K. Duerr R.H. McGovern D.P. Hui K.Y. Lee J.C. Schumm L.P. Sharma Y. Anderson C.A. et al.International IBD Genetics Consortium (IIBDGC)Host-microbe interactions have shaped the genetic architecture of inflammatory bowel disease.Nature. 2012; 491: 119-124Crossref PubMed Scopus (3230) Google Scholar), and roles for adipocyte thermogenesis (Claussnitzer et al., 2015Claussnitzer M. Dankel S.N. Kim K.H. Quon G. Meuleman W. Haugen C. Glunk V. Sousa I.S. Beaudry J.L. Puviindran V. et al.FTO Obesity Variant Circuitry and Adipocyte Browning in Humans.N. Engl. J. Med. 2015; 373: 895-907Crossref PubMed Scopus (799) Google Scholar) and central nervous system genes in obesity (Locke et al., 2015Locke A.E. Kahali B. Berndt S.I. Justice A.E. Pers T.H. Day F.R. Powell C. Vedantam S. Buchkovich M.L. Yang J. et al.LifeLines Cohort StudyADIPOGen ConsortiumAGEN-BMI Working GroupCARDIOGRAMplusC4D ConsortiumCKDGen ConsortiumGLGCICBPMAGIC InvestigatorsMuTHER ConsortiumMIGen ConsortiumPAGE ConsortiumReproGen ConsortiumGENIE ConsortiumInternational Endogene ConsortiumGenetic studies of body mass index yield new insights for obesity biology.Nature. 2015; 518: 197-206Crossref PubMed Scopus (2645) Google Scholar). But despite the success of these earlier studies, we argue that the enrichment of signal in relevant genes is surprisingly weak overall, suggesting that prevailing conceptual models for complex diseases are incomplete. We highlight some pertinent features of current data and discuss what these may tell us about the genetic architecture of complex diseases. Early practitioners of GWAS were dismayed to find that, for most traits, the strongest genetic associations could explain only a small fraction of the genetic variance (Manolio et al., 2009Manolio T.A. Collins F.S. Cox N.J. Goldstein D.B. Hindorff L.A. Hunter D.J. McCarthy M.I. Ramos E.M. Cardon L.R. Chakravarti A. et al.Finding the missing heritability of complex diseases.Nature. 2009; 461: 747-753Crossref PubMed Scopus (5883) Google Scholar). This was taken to imply that there must be many causal loci, each with small effect sizes (Goldstein, 2009Goldstein D.B. Common genetic variation and human traits.N. Engl. J. Med. 2009; 360: 1696-1698Crossref PubMed Scopus (749) Google Scholar). Subsequent analyses soon provided direct evidence for this in the case of schizophrenia (Purcell et al., 2009Purcell S.M. Wray N.R. Stone J.L. Visscher P.M. O'Donovan M.C. Sullivan P.F. Sklar P. Ruderfer D.M. McQuillin A. Morris D.W. et al.International Schizophrenia ConsortiumCommon polygenic variation contributes to risk of schizophrenia and bipolar disorder.Nature. 2009; 460: 748-752Crossref PubMed Scopus (3442) Google Scholar) and showed that, together, common variants can explain much of the expected heritability (Yang et al., 2010Yang J. Benyamin B. McEvoy B.P. Gordon S. Henders A.K. Nyholt D.R. Madden P.A. Heath A.C. Martin N.G. Montgomery G.W. et al.Common SNPs explain a large proportion of the heritability for human height.Nat. Genet. 2010; 42: 565-569Crossref PubMed Scopus (2793) Google Scholar). While traits vary greatly in terms of both the importance of the largest-effect common variants and of higher-penetrance rare variants (Loh et al., 2015Loh P.-R. Bhatia G. Gusev A. Finucane H.K. Bulik-Sullivan B.K. Pollack S.J. de Candia T.R. Lee S.H. Wray N.R. Kendler K.S. et al.Schizophrenia Working Group of Psychiatric Genomics ConsortiumContrasting genetic architectures of schizophrenia and other complex diseases using fast variance-components analysis.Nat. Genet. 2015; 47: 1385-1392Crossref PubMed Scopus (245) Google Scholar, Shi et al., 2016Shi H. Kichaev G. Pasaniuc B. Contrasting the genetic architecture of 30 complex traits from summary association data.Am. J. Hum. Genet. 2016; 99: 139-153Abstract Full Text Full Text PDF PubMed Scopus (183) Google Scholar, Sullivan et al., 2017Sullivan P.F. Agrawal A. Bulik C. Andreassen O.A. Borglum A. Breen G. Cichon S. Edenberg H. Faraone S.V. Gelernter J. Mathews C.A. Nievergelt C.M. Smoller J. O'Donovan M. Psychiatric Genomics: An Update and an Agenda.bioRxiv. 2017; https://doi.org/10.1101/115600Google Scholar), it is now clear that polygenic effects are important across a wide variety of traits (Shi et al., 2016Shi H. Kichaev G. Pasaniuc B. Contrasting the genetic architecture of 30 complex traits from summary association data.Am. J. Hum. Genet. 2016; 99: 139-153Abstract Full Text Full Text PDF PubMed Scopus (183) Google Scholar, Weiner et al., 2016Weiner D.J. Wigdor E.M. Ripke S. Walters R.K. Kosmicki J.A. Grove J. Samocha K.E. Goldstein J. Okbay A. Bybjerg-Gauholm J. et al.Polygenic transmission disequilibrium confirms that common and rare variation act additively to create risk for autism spectrum disorders.bioRxiv. 2016; https://doi.org/10.1101/089342Google Scholar). One key question that has been under-studied to date is the extent to which causal variants are spread widely across the genome or clumped into disease-relevant pathways. However, it is known that the heritability contributed by each chromosome tends to be closely proportional to its physical length (Visscher et al., 2006Visscher P.M. Medland S.E. Ferreira M.A. Morley K.I. Zhu G. Cornes B.K. Montgomery G.W. Martin N.G. Assumption-free estimation of heritability from genome-wide identity-by-descent sharing between full siblings.PLoS Genet. 2006; 2: e41Crossref PubMed Scopus (396) Google Scholar, Shi et al., 2016Shi H. Kichaev G. Pasaniuc B. Contrasting the genetic architecture of 30 complex traits from summary association data.Am. J. Hum. Genet. 2016; 99: 139-153Abstract Full Text Full Text PDF PubMed Scopus (183) Google Scholar), hinting that causal variants may be fairly uniformly distributed. And recent data show that causal variants can be surprisingly dispersed even at finer scales. A paper from Alkes Price and colleagues estimated that 71%–100% of 1-MB windows in the genome contribute to heritability for schizophrenia (Loh et al., 2015Loh P.-R. Bhatia G. Gusev A. Finucane H.K. Bulik-Sullivan B.K. Pollack S.J. de Candia T.R. Lee S.H. Wray N.R. Kendler K.S. et al.Schizophrenia Working Group of Psychiatric Genomics ConsortiumContrasting genetic architectures of schizophrenia and other complex diseases using fast variance-components analysis.Nat. Genet. 2015; 47: 1385-1392Crossref PubMed Scopus (245) Google Scholar). Here we explore a second example—namely, height—for which very large GWAS datasets are available (Figure 1). While height is often thought of as the quintessential polygenic trait, recent work shows that the genetic architecture of height is actually broadly similar to that of a wide variety of other quantitative traits and diseases ranging from diabetes or autoimmune diseases to BMI or cholesterol levels. Thus, we use height to illustrate the extreme polygenicity typical of many complex traits (Shi et al., 2016Shi H. Kichaev G. Pasaniuc B. Contrasting the genetic architecture of 30 complex traits from summary association data.Am. J. Hum. Genet. 2016; 99: 139-153Abstract Full Text Full Text PDF PubMed Scopus (183) Google Scholar, Chakravarti and Turner, 2016Chakravarti A. Turner T.N. Revealing rate-limiting steps in complex disease biology: The crucial importance of studying rare, extreme-phenotype families.BioEssays. 2016; 38: 578-586Crossref PubMed Scopus (33) Google Scholar). A height meta-analysis from the GIANT study reported 697 genome-wide significant loci that, together, explain 16% of the phenotypic variance (Wood et al., 2014Wood A.R. Esko T. Yang J. Vedantam S. Pers T.H. Gustafsson S. Chu A.Y. Estrada K. Luan J. Kutalik Z. et al.Electronic Medical Records and Genomics (eMEMERGEGE) ConsortiumMIGen ConsortiumPAGEGE ConsortiumLifeLines Cohort StudyDefining the role of common variation in the genomic and biological architecture of adult human height.Nat. Genet. 2014; 46: 1173-1186Crossref PubMed Scopus (1198) Google Scholar). But a quantile-quantile plot comparing the distribution of p values against the expected null distribution shows that the distribution of p values is hugely shifted toward small p values (Figure 1A), such that common variants together explain 86% of the expected heritability (Shi et al., 2016Shi H. Kichaev G. Pasaniuc B. Contrasting the genetic architecture of 30 complex traits from summary association data.Am. J. Hum. Genet. 2016; 99: 139-153Abstract Full Text Full Text PDF PubMed Scopus (183) Google Scholar). The inflation is stronger in active chromatin and in expression quantitative trait loci (eQTLs), consistent with the expected enrichment of signal in gene-regulatory regions. We next used ashR to analyze the distribution of regression coefficients from the set of all SNPs (Stephens, 2017Stephens M. False discovery rates: a new deal.Biostatistics. 2017; 18: 275-294PubMed Google Scholar). ashR models the GWAS results as a mixture of SNPs that have a true effect size of exactly zero, with SNPs that have a true effect size that is not zero. Using this approach, we estimated that, remarkably, 62% of all common SNPs are associated with a non-zero effect on height (this includes both causal SNPs as well as nearby SNPs that are correlated through linkage disequilibrium; Figure 1B). Given that the typical extent of linkage disequilibrium (LD) is around 10–100 kb (International HapMap Consortium, 2005International HapMap ConsortiumA haplotype map of the human genome.Nature. 2005; 437: 1299-1320Crossref PubMed Scopus (4773) Google Scholar), this implies that most 100-kb windows in the genome include variants that affect height. Stratifying the ashR analysis by the LD score for each SNP (Bulik-Sullivan et al., 2015bBulik-Sullivan B.K. Loh P.R. Finucane H.K. Ripke S. Yang J. Patterson N. Daly M.J. Price A.L. Neale B.M. Schizophrenia Working Group of the Psychiatric Genomics ConsortiumLD Score regression distinguishes confounding from polygenicity in genome-wide association studies.Nat. Genet. 2015; 47: 291-295Crossref PubMed Scopus (1923) Google Scholar), we see a clear effect that SNPs with more LD partners are more likely to be associated with height. Under simplifying assumptions (see Supplemental Information), the best-fit curve suggests that ∼3.8% of 1000 Genomes SNPs have causal effects on height. As validation, we used the regression estimate from each SNP in the height meta-analysis to predict its direction of effect on height (Figure 1C) and then examined the extent to which SNP effects are consistent in a smaller, independent dataset from the Health and Retirement Study (Juster and Suzman, 1995Juster F.T. Suzman R. An overview of the Health and Retirement Study.J. Hum. Resour. 1995; 30: S7-S56Crossref Google Scholar). In brief, we computed the mean replication effect sizes of height-increasing alleles as determined by GIANT. Under the null hypothesis of no true signal, the replication effect sizes would be centered on zero; when there is true signal, the observed mean effect sizes can be considered a lower bound on the true effect sizes due to occasional sign errors in GIANT. Strikingly, we find clear enrichment of shared directional signal for most SNPs, even for SNPs with p values as large as 0.5 (Figure 1C). Across all SNPs genome-wide, the median SNP is associated with an effect size of 0.14 mm, which is approximately one-tenth the median effect size of genome-wide significant SNPs (1.43 mm). We also obtained similar results starting from a smaller family-based GWAS, confirming that the signals are not driven by confounding from population structure (Supplemental Information). Putting the various lines of evidence together, we estimate that more than 100,000 SNPs exert independent causal effects on height, similar to an early estimate of 93,000 causal variants based on a different approach (Goldstein, 2009Goldstein D.B. Common genetic variation and human traits.N. Engl. J. Med. 2009; 360: 1696-1698Crossref PubMed Scopus (749) Google Scholar) (Supplemental Information). In summary, we conclude that there is an extremely large number of causal variants with tiny effect sizes on height and, moreover, that these are spread very widely across the genome, such that most 100-kb windows contribute to variance in height. More generally, the heritability of complex traits and diseases is spread broadly across the genome (Loh et al., 2015Loh P.-R. Bhatia G. Gusev A. Finucane H.K. Bulik-Sullivan B.K. Pollack S.J. de Candia T.R. Lee S.H. Wray N.R. Kendler K.S. et al.Schizophrenia Working Group of Psychiatric Genomics ConsortiumContrasting genetic architectures of schizophrenia and other complex diseases using fast variance-components analysis.Nat. Genet. 2015; 47: 1385-1392Crossref PubMed Scopus (245) Google Scholar, Shi et al., 2016Shi H. Kichaev G. Pasaniuc B. Contrasting the genetic architecture of 30 complex traits from summary association data.Am. J. Hum. Genet. 2016; 99: 139-153Abstract Full Text Full Text PDF PubMed Scopus (183) Google Scholar), implying that a substantial fraction of all genes contribute to variation in disease risk. These observations seem inconsistent with the expectation that complex trait variants are primarily in specific biologically relevant genes and pathways. To explore this further, we turn next to data on functional enrichment of signals. As shown above for height, GWAS signals tend to be markedly enriched in predicted gene regulatory elements. In particular, many groups have shown that disease-associated SNPs are enriched in active chromatin and particularly in chromatin that is active in cell types relevant to disease (Trynka et al., 2013Trynka G. Sandor C. Han B. Xu H. Stranger B.E. Liu X.S. Raychaudhuri S. Chromatin marks identify critical cell types for fine mapping complex trait variants.Nat. Genet. 2013; 45: 124-130Crossref PubMed Scopus (385) Google Scholar, Farh et al., 2015Farh K.K.-H. Marson A. Zhu J. Kleinewietfeld M. Housley W.J. Beik S. Shoresh N. Whitton H. Ryan R.J. Shishkin A.A. et al.Genetic and epigenetic fine mapping of causal autoimmune disease variants.Nature. 2015; 518: 337-343Crossref PubMed Scopus (1153) Google Scholar, Finucane et al., 2015Finucane H.K. Bulik-Sullivan B. Gusev A. Trynka G. Reshef Y. Loh P.-R. Anttila V. Xu H. Zang C. Farh K. et al.ReproGen ConsortiumSchizophrenia Working Group of the Psychiatric Genomics ConsortiumRACI ConsortiumPartitioning heritability by functional annotation using genome-wide association summary statistics.Nat. Genet. 2015; 47: 1228-1235Crossref PubMed Scopus (980) Google Scholar, Kundaje et al., 2015Kundaje A. Meuleman W. Ernst J. Bilenky M. Yen A. Heravi-Moussavi A. Kheradpour P. Zhang Z. Wang J. Ziller M.J. et al.Roadmap Epigenomics ConsortiumIntegrative analysis of 111 reference human epigenomes.Nature. 2015; 518: 317-330Crossref PubMed Scopus (3578) Google Scholar). Similarly, signals also aggregate near genes that are expressed in relevant cell types (Hu et al., 2011Hu X. Kim H. Stahl E. Plenge R. Daly M. Raychaudhuri S. Integrating autoimmune risk loci with gene-expression data identifies specific pathogenic immune cell subsets.Am. J. Hum. Genet. 2011; 89: 496-506Abstract Full Text Full Text PDF PubMed Scopus (123) Google Scholar, Wood et al., 2014Wood A.R. Esko T. Yang J. Vedantam S. Pers T.H. Gustafsson S. Chu A.Y. Estrada K. Luan J. Kutalik Z. et al.Electronic Medical Records and Genomics (eMEMERGEGE) ConsortiumMIGen ConsortiumPAGEGE ConsortiumLifeLines Cohort StudyDefining the role of common variation in the genomic and biological architecture of adult human height.Nat. Genet. 2014; 46: 1173-1186Crossref PubMed Scopus (1198) Google Scholar). An intuitive interpretation is that the cell-type-based regulatory maps point us toward cell-type-specific regulatory elements that control specific functions of those cells and thereby drive disease biology. Indeed, the relevant papers often describe these analyses as highlighting "cell-type-specific" aspects of regulation. But given that the heritability signal is so widespread, we wanted to understand whether the signal is specifically concentrated in chromatin that is active in just the relevant (or related) cell types, as opposed to chromatin that is broadly active. To explore this question, we used active chromatin data measured in ten broadly defined cell-type groups (e.g., immune, central nervous system (CNS), cardiovascular, etc.). A region was considered active in a cell-type group if it was detected as active for any cell type in that group. We applied stratified LD score regression—a method that estimates how much different classes of SNPs contribute to heritability (Finucane et al., 2015Finucane H.K. Bulik-Sullivan B. Gusev A. Trynka G. Reshef Y. Loh P.-R. Anttila V. Xu H. Zang C. Farh K. et al.ReproGen ConsortiumSchizophrenia Working Group of the Psychiatric Genomics ConsortiumRACI ConsortiumPartitioning heritability by functional annotation using genome-wide association summary statistics.Nat. Genet. 2015; 47: 1228-1235Crossref PubMed Scopus (980) Google Scholar). We focused on three well-powered GWAS studies that showed clear enrichment within a single cell-type group in a previous analysis: Crohn's disease (immune), rheumatoid arthritis (RA, immune), and schizophrenia (CNS) (Finucane et al., 2015Finucane H.K. Bulik-Sullivan B. Gusev A. Trynka G. Reshef Y. Loh P.-R. Anttila V. Xu H.