The missing link between genetic association and regulatory function

Department of Biomedical Informatics, Harvard Medical School, United States
Brigham and Women’s Hospital, Division of Genetics, Harvard Medical School, United States
Program in Medical and Population Genetics, Broad Institute of MIT and Harvard, United States
Brigham and Women’s Hospital, Department of Neurology, Harvard Medical School, United States
Department of Epidemiology, Harvard T.H. Chan School of Public Health, United States
Altius Institute, United States
Division of Pulmonary Medicine, Boston Children’s Hospital, United States
Department of Neurology, Yale Medical School, United States
Department of Genetics, Yale Medical School, United States

Dec 14, 2022

https://doi.org/10.7554/eLife.74970

Open access
Copyright information

Version of Record

Accepted for publication after peer review and revision.

Download
Cite
Share
CommentOpen annotations (there are currently 0 annotations on this page).

Version of Record published: January 16, 2023 (This version)
Accepted Manuscript published: December 14, 2022 (Go to version)
Accepted: December 2, 2022
Received: October 25, 2021
Preprint posted: June 11, 2021 (Go to version)

Figures
Tables
Additional files

3 figures, 4 tables and 4 additional files

Figures

Figure 1 with 2 supplements

Download asset Open asset

Putatively causative genes identified by each method category.

The leftmost column in each half of the plot displays the entire group of putatively causative genes for our Mendelian set of genes and our (Backman et al., 2021) set of genes, respectively, as well as noting how many are unique to each set or shared between the two sets. The second column in each half indicates how many genes from each set have a nearby GWAS peak or have both a nearby GWAS peak and an expression QTL (eQTL). The remaining columns indicate how many genes were identified through colocalization, transcriptome-wide association studies (TWAS), or chromatin methods, while noting how many of these genes are unique vs. shared between the Mendelian and Backman sets.

Figure 1—figure supplement 1

Download asset Open asset

Enrichment of Mendelian genes near GWAS peaks.

(A) As the window around GWAS peaks shrinks, the enrichment of Mendelian genes within the window becomes increasingly significant, while the enrichment of non-matching trait pairs used as controls (gray lines; permutation test described in 'Materials and methods') is not consistently increased. Some controls achieve nominal significance (dotted horizontal line), but none reach significance once multiple testing is corrected for (solid horizontal line). (B) As above, but for genes from Backman et al., 2021. (C) The combined gene lists from parts (A) and (B). Note that, accounting for multiple test correction (based on the total number of tests across all panels), height does not reach significance using the Mendelian gene list, while T2D is barely significant using the Backman list. However, combining the lists increases power and demonstrates significance for all traits.

Figure 1—figure supplement 2

Download asset Open asset

Change in coloc hits when adjusting expression QTL(eQTL) statistics using Multivariate Adaptive Shrinkage Method (MASH).

By using the Bayesian method MASH to update our measurements of eQTLs based on tissues with similar expression patterns, we increased the number of colocalizations found. However, even in tissues in which the number of genes identified increased substantially, we did not meaningfully increase the number of putatively causative genes identified.

Figure 2

Download asset Open asset

Genes identified as associated with a complex trait by each method.

Columns 'Mend' and 'Backman' indicate whether a gene is from the Mendelian set of putatively causative genes, the Backman et al. set, or both. Subsequent columns indicate whether a gene was identified as a hit using each of our methods: JLIM, coloc, eCaviar, transcriptome-wide association studies (TWAS), and chromatin analysis.

Figure 3

Download asset Open asset

Chromatin-based causative gene identification.

Following the fine-mapping of GWAS variants, three parallel methods were used. The first identified fine-mapped variants falling within regions annotated as enhancers by ChromHMM. The second identified variants within histone modification features and evaluated their relevance using an activity-by-distance (ABD) score that combined the strength of the feature (i.e., the strength of the acetylation or methylation peak) with its genomic distance to the gene of interest ('Materials and methods'). The third repeated both of these—checking for fine-mapped variants within a region and calculating the ABD score—for DNase I hypersensitivity sites.

Figure 3—source data 1 Gene-level results for linked expression and traits.: https://cdn.elifesciences.org/articles/74970/elife-74970-fig3-data1-v2.txt
Download elife-74970-fig3-data1-v2.txt

Tables

Table 1

Putatively causative Mendelian genes.

Each gene includes reference(s) to the known biological role of its coding variants, as established in familial studies, in vitro experiments, and/or animal models. Genes from Backman et al., 2021 are not included here, but can be found in Figure 2.

Phenotype	Genes
Low-density lipoprotein	APOB Soria et al., 1989; Pullinger et al., 1995 APOC2 Hegele et al., 1991 APOE de Knijff et al., 1994 LDLR Brown and Goldstein, 1976 LPL Heizmann et al., 1991; Clee et al., 2001 PCSK9 Abifadel et al., 2003
High-density lipoprotein	ABCA1 Brooks-Wilson et al., 1999; Bodzioch et al., 1999; Rust et al., 1999; Ordovas et al., 1986 APOA1 Ordovas et al., 1986 CETP Glueck et al., 1975 LIPC Isaacs et al., 2004; Grarup et al., 2008; Iijima et al., 2008 LIPG Yamakawa-Kobayashi et al., 2003 PLTP Jiang et al., 1999 SCARB1 Tai et al., 2003; McCarthy et al., 2003
Height	ANTXR1 Stránecký et al., 2013; Bayram et al., 2014 ATR O’Driscoll et al., 2003; Ogi et al., 2012 BLM Ellis et al., 1995; Foucault et al., 1997 CDC6 Bicknell et al., 2011a CDT1 Bicknell et al., 2011a; Guernsey et al., 2011 CENPJ AlDosari et al., 2010 COL1A1 Wallis et al., 1990 COL1A2 Spotila et al., 1992; De Paepe et al., 1997 COMP Briggs et al., 1995; Mabuchi et al., 2003 CREBBP Menke et al., 2016; Menke et al., 2018; Angius et al., 2019 DNA2 Shaheen et al., 2014 EP300 Woods et al., 2014; Tsai et al., 2011 EVC Polymeropoulos et al., 1996; Ruiz-Perez et al., 2003 EVC2 Ruiz-Perez et al., 2003; Galdzicka et al., 2002 BN1 Faivre et al., 2003; Le Goff et al., 2011; Horn and Robinson, 2011; Takenouchi et al., 2013 FGFR3 Hyland et al., 2003; Toydemir et al., 2006; Makrythanasis et al., 2014 FKBP10 Alanay et al., 2010; Kelley et al., 2011; Barnes et al., 2013 HR Berg et al., 1993; Woods et al., 1996; Goddard et al., 1995; Ayling et al., 1997 KRAS Aoki et al., 2005; Schubbert et al., 2006; Carta et al., 2006 NBN Varon et al., 1998; Tanzanella et al., 2003 NIPBL Tonkin et al., 2004; Krantz et al., 2004 ORC1 Bicknell et al., 2011a; Guernsey et al., 2011; Bicknell et al., 2011b RC4 Guernsey et al., 2011; Bicknell et al., 2011b ORC6L Bicknell et al., 2011a; de Munnik et al., 2012 PCNT Rauch et al., 2008; Griffith et al., 2008; Piane et al., 2009 PLOD2 van der Slot et al., 2003; HaVinh et al., 2004; Puig-Hervás et al., 2012 PTPN11 Tartaglia et al., 2001; Maheshwari et al., 2002; Kosaki et al., 2002 RAD21 Deardorff et al., 2012; Kruszka et al., 2019; Goel and Parasivam, 2020 RAF1 Pandit et al., 2007; Razzaque et al., 2007 RECQL4 Lindor et al., 2000; Beghini et al., 2003; Wang et al., 2003 RIT1 Aoki et al., 2013; Bertola et al., 2014; Gos et al., 2014 ROR2 Afzal et al., 2000; van Bokhoven, 2000; Tufan et al., 2005 SLC26A2 Hästbacka et al., 1993; Rossi and Superti-Furga, 2001; Barreda-Bonis et al., 2018 SMAD4 Le Goff et al., 2012; Caputo et al., 2012; Lindor et al., 2012 SRCAP Hood et al., 2012; Le Goff et al., 2013 WRN Yu et al., 1996; Goto et al., 1997; Yu, 1997
Crohn disease	ATG16L1 Hampe et al., 2007 CARD9 Rivas et al., 2011 IL10 Fowler, 2005 IL10RA Gasche et al., 2003; Mao et al., 2012 IL10RB Glocker et al., 2009; Begue et al., 2011 IL23R Duerr et al., 2006; Libioulle et al., 2007; Glas et al., 2007 IRGM McCarroll et al., 2008; Craddock et al., 2010; Prescott et al., 2010 NOD2 Ogura et al., 2001; Hugot et al., 2001 PRDM1 Ellinghaus et al., 2013 PTPN22 Diaz-Gallo et al., 2011
Ulcerative colitis	ATG16L1 Fowler et al., 2008 CARD9 Rivas et al., 2011 IL23R Glas et al., 2007; Fisher et al., 2008 IRGM McCarroll et al., 2008 PRDM1 Ellinghaus et al., 2013 PTPN22 Diaz-Gallo et al., 2011 RNF186 Rivas et al., 2016; Beaudoin et al., 2013
Type II diabetes	ABCC8 Reis et al., 2000 BLK Borowiec et al., 2009 CEL Bengtsson-Ellmark et al., 2004; Raeder et al., 2006 EIF2AK3 Harding et al., 2001; Brickwood et al., 2003; Durocher et al., 2006 GATA4 Shaw-Smith, 2014 GATA6 Yorifuji et al., 2012; De Franco et al., 2013 GCK Froguel et al., 1993 GLIS3 Senée et al., 2006 HNF1A Yamagata et al., 1996b; Vaxillaire et al., 1997 HNF1B Horikawa et al., 1997; Lindner et al., 1999 HNF4A Yamagata et al., 1996a; Stoffel and Duncan, 1997 IER3IP1 Poulton et al., 2011; Abdel-Salam et al., 2012; Shalev et al., 2014 INS Støy et al., 2007 KCNJ11 Hani et al., 1998; Gloyn et al., 2004 j KLF11 Neve et al., 2005 LMNA Cao and Hegele, 2000 NEUROD1 Malecki et al., 1999 NEUROG3 Gradwohl et al., 2000; Rubio-Cabezas et al., 2011; Pinney et al., 2011 PAX4 Shimajiri et al., 2001; Mauvais-Jarvis et al., 2004; Plengvidhya et al., 2007 PDX1 Stoffers et al., 1997; Macfarlane et al., 1999; Hani et al., 1999 PPARG Deeb et al., 1998; Savage et al., 2002 PTF1A Sellick et al., 2004 RFX6 Smith, 2010; Sansbury et al., 2015 SLC19A2 Labay et al., 1999, Oishi et al., 2002; Shaw-Smith et al., 2012 SLC2A2 Laukkanen et al., 2005; Sansbury et al., 2012 WFS1 Strom et al., 1998; Hardy et al., 1999; Khanim et al., 2001 ZFP57 Mackay et al., 2008; Boonen et al., 2013
Breast cancer (selected using MutPanning; Dietlein et al., 2020)	AKT1 ARID1A ATM BRCA1 BRCA2 CBFB CDH1 CDKN1B CHEK2 CTCF ERBB2 ESR1 FGFR2 FOXA1 GATA3 GPS2 HS6ST1 KMT2C KRAS LRRC37A3 MAP2K4 MAP3K1 NCOR1 NF1 NUP93 PALB2 PIK3CA PTEN RB1 RUNX1 SF3B1 STK11 TBX3 TP53 ZFP36L1

Table 2

Tissue-trait pairs.

Tissues were selected for each trait based on a priori knowledge of disease biology.

Mendelian trait	GWAS trait	Tissues examined
Breast cancer	Breast cancer	Breast mammary tissue
Crohn disease	Crohn disease	Small intestine terminal ileum Colon sigmoid Colon transverse
Ulcerative colitis	Ulcerative colitis	Small intestine terminal ileum Colon sigmoid Colon transverse
Dyslipidemia Hyperlipidemia Tangier’s disease	High-density lipoprotein	Liver Adipose (subcutaneous) Whole blood
Dyslipidemia Hyperlipidemia	Low-density lipoprotein	Liver Adipose (subcutaneous) Whole blood
Mendelian short stature	Height	Skeletal muscle
Monogenic diabetes	Type II diabetes	Pancreas Skeletal muscle Adipose (subcutaneous) Small intestine terminal ileum

Table 3

Proposed explanations for negative results under the unembellished model.

Many explanations have been proposed for GWAS associations that are not explained by cis-QTLs. This table details the explanations inconsistent with our results, which are explained in the left column and addressed on the right. Explanations involving more detailed models of gene regulation can be found in Table 4. Two of the explanations addressed here involve violations of the assumptions of our and other expression-based complex trait studies. If coding and non-coding variants affect fundamentally different biological pathways, or if trait associations rarely depend on cis-eQTLs, our methods of mapping regulation to traits would have nothing to uncover. Even in the presence of eQTL-driven trait associations, insufficient power to detect trait associations, to detect eQTL associations, or to link the two would result in predominantly negative results.

Violated assumptions
Genes implicated via coding variants are irrelevant for non-coding associations	Our genes are enriched for GWAS associations even after removing the effects of coding variants Loss-of-function variants, which underlie many Mendelian-trait genes, can be thought of as large-effect eQTLs Genes identified from Backman et al., 2021 are not based on cognate phenotypes, but the same complex phenotypes as GWAS
Regulatory mechanisms other than cis-eQTLs	Splice QTLs are consistently found to explain less phenotypic variance than eQTLs, and they cannot explain the many GWAS associations that fall within intergenic regions Trans-eQTLs are believed to rely on their effects as cis-eQTLs for other genes; the few exceptions to this model (e.g., CTCF binding sites) are not broadly applicable
Insufficient power
Lack of GWAS power	GWAS have been shown to have sufficient power to identify small effects even in rare variants 2/3 of the genes we used have nearby GWAS associations, reflecting a strong enrichment and indicating that GWAS discovery is not a limiting factor Our analysis is conditioned on the presence of GWAS associations
Lack of eQTL mapping power	GTEx is well powered for eQTL discovery in bulk tissue GTEx Consortium, 2020; Gamazon et al., 2015 93% of our genes have a mapped cis-eQTL in a relevant tissue
Lack of power for colocalization and TWAS methods	Simulations show that colocalization and TWAS methods are well-powered Chun et al., 2017; Giambartolomei et al., 2018; Hormozdiari et al., 2016; Gusev et al., 2016; Hukku et al., 2021 They are robust to levels of LD mismatch higher than what would be expected given our datasets Chun et al., 2017; Wainberg et al., 2019; Hukku et al., 2021 Some, though not all, of the methods are robust to allelic heterogeneity Chun et al., 2017; Wainberg et al., 2019

eQTL = expression QTL; TWAS = transcriptome-wide association studies.

Table 4

Explaining negative results with more nuanced models of gene regulation.

To reconcile an expression-based model with our observations requires us to both explain the absence of trait-linked eQTLs as well as explaining away the inconsequence of eQTLs for trait-linked genes. The left-hand side lists additions or changes to the unembellished model, while the right-hand side contains explanations of the models and current relevant research.

Extended models of gene regulation
Context dependency: a context-specific eQTL, invisible in bulk tissues analyzed to date, replaces or supplements the bulk tissue homeostatic eQTL	Cell type Dobbyn et al., 2018; Zhang et al., 2018; Schmiedel et al., 2018; Glastonbury et al., 2019; Rai et al., 2020; Findley et al., 2021; Neavin et al., 2021; Ota et al., 2021; Patel et al., 2021; Bryois et al., 2021; Arvanitis et al., 2022; Oelen et al., 2022; Perez et al., 2022; Schmiedel et al., 2022; Yazar et al., 2022 Only a subset of cell types in the tissue contribute to the GWAS phenotype. An eQTL specific to such a cell type is causative for the phenotype. The eQTL either cannot be detected in bulk tissue because of the cell type’s low prevalence The appropriate cellular or anatomical context has not yet been analyzed
	Developmental timing Dobbyn et al., 2018; Strober et al., 2019; Cuomo et al., 2020; Bonder et al., 2021; Jerber et al., 2021; Aygün et al., 2022; Elorbany et al., 2022 The GWAS phenotype depends on a specific point in cell/tissue development or differentiation eQTLs present at the correct interval contribute to phenotype, but eQTLs observed at other points do not
	Cell state or environment Findley et al., 2021; Ota et al., 2021; Oelen et al., 2022; Schmiedel et al., 2022; Huh and Paulsson, 2011; Knowles et al., 2017; Kim-Hellmuth et al., 2017; Balliu et al., 2021; Mu et al., 2021; Ward et al., 2021; Nathan et al., 2022; Baca et al., 2022 The causative eQTL has effects that are undetectable in steady-state expression under normal conditions It may activate only in response to a specific environmental condition, such as immune activation or a metabolic shift
Nonlinear or non-homeostatic: the relationship between eQTL and genotype is indirect	Nonlinearity Fu et al., 2009; Dori-Bachash et al., 2011; Ghazalpour et al., 2011; Pai et al., 2012; Vogel and Marcotte, 2012; Khan et al., 2013; Wu et al., 2013; McManus et al., 2014; Albert and Kruglyak, 2015; Bader et al., 2015; Battle et al., 2015; Cenik et al., 2015; McManus et al., 2015; Pai et al., 2015; Schafer et al., 2015; Chick et al., 2016; Liu et al., 2016; Schaefke et al., 2018; Buccitelli and Selbach, 2020; Wang et al., 2020a; Kusnadi et al., 2022 There may be buffering that prevents a change in expression from producing a change in protein levels Expression below a certain level may not influence phenotype, rendering small eQTLs irrelevant
	Steady-state expression may be a poor model Pedraza and Paulsson, 2008; Raj and van Oudenaarden, 2008; Shahrezaei and Swain, 2008; Larson et al., 2009; Raj and van Oudenaarden, 2009; Suter et al., 2011; Dar et al., 2012; Viñuelas et al., 2013; Kumar et al., 2015; Nicolas et al., 2017; Qiu et al., 2019; Wang et al., 2020c Phenotype may depend on the kinetics of expression, which could be cyclical or follow some other pattern Expression may be stochastic, such that only a random subset of cells display the relevant expression pattern at any one time