Meta-analysis fine-mapping is often miscalibrated at single-variant resolution

Kanai, Masahiro; Elzur, Roy; Zhou, Wei; Wu, Kuan-Han Hank; Rasheed, Humaira; Tsuo, Kristin; Hirbo, Jibril; Wang, Ying; Bhattacharya, Arjun; Zhao, Huiling; Namba, Shinichi; Surakka, Ida; Wolford, Brooke N.; Faro, Valeria Lo; Lopera-Maya, Esteban A.; Läll, Kristi; Favé, Marie-Julie; Partanen, Juulia; Chapman, Sinéad B.; Karjalainen, Juha; Kurki, Mitja; Maasha, Mutaamba; Brumpton, Ben; Chavan, Sameer; Chen, Tzu-Ting; Daya, Michelle; Ding, Yi; Feng, Yen‐Chen Anne; Guare, Lindsay; Gignoux, Christopher R.; Graham, Sarah E.; Hornsby, Whitney E.; Ingold, Nathan; Ismail, Said I.; Johnson, Ruth; Laisk, Triin; Lin, Kuang; Lv, Jun; Millwood, Iona Y.; Moreno‐Grau, Sonia; Nam, Kisung; Palta, Priit; Pandit, Anita; Preuss, Michael; Saad, Chadi; Setia-Verma, Shefali; Þorsteinsdóttir, Unnur; Uzunović, Jasmina; Verma, Anurag; Zawistowski, Matthew; Zhong, Xue; Afifi, Nahla; Al-Dabhani, Kawthar; Thani, Asma Al; Bradford, Yuki; Campbell, Archie; Crooks, Kristy; Bock, Geertruida H. de; Damrauer, Scott M.; Douville, Nicholas J.; Finer, Sarah; Fritsche, Lars G.; Fthenou, Eleni; Gonzalez-Arroyo, Gilberto; Griffiths, Christopher J.; Guo, Yu; Hunt, Karen A.; Ioannidis, Alexander; Jansonius, Nomdo M.; Konuma, Takahiro; Lee, Ming Ta Michael; Lopez-Pineda, Arturo; Matsuda, Yuta; Marioni, Riccardo E.; Moatamed, Babak; Nava-Aguilar, Marco A.; Numakura, Kensuke; Patil, Snehal; Rafaels, Nicholas; Richmond, Anne; Rojas-Muñoz, Agustin; Shortt, Jonathan A.; Straub, Péter; Tao, Ran; Vanderwerff, Brett; Vernekar, Manvi; Veturi, Yogasudha; Barnes, Kathleen C.; Boezen, Marike; Chen, Zhengming; Chen, Chia‐Yen; Cho, Judy H.; Smith, George Davey; Finucane, Hilary; Franke, Lude; Gamazon, Eric R.; Ganna, Andrea; Gaunt, Tom R.; Ge, Tian; Huang, Hailiang; Huffman, Jennifer E.; Katsanis, Nicholas; Koskela, Jukka; Lajonchere, Clara; Law, Matthew; Li, Liming; Lindgren, Cecilia M.; Loos, Ruth J. F.; MacGregor, Stuart; Matsuda, Koichi; Olsen, Catherine M.; Porteous, David J.; Shavit, Jordan A.; Snieder, Harold; Takano, Toru; Trembath, Richard; Vonk, Judith M.; Whiteman, David C.; Wicks, Stephen J.; Wijmenga, Cisca; Wright, John; Zheng, Jie; Zhou, Xiang; Awadalla, Philip; Boehnke, Michael; Bustamante, Carlos D.; Cox, Nancy J.; Fatumo, Segun; Geschwind, Daniel H.; Hayward, Caroline; Hveem, Kristian; Kenny, Eimear E.; Lee, Seunggeun; Lin, Yen-Feng; Mbarek, Hamdi; Mägi, Reedik; Martin, Hilary C.; Medland, Sarah E.; Okada, Yukinori; Palotie, Aarno; Paşaniuc, Bogdan; Rader, Daniel J.; Ritchie, Marylyn D.; Sanna, Serena; Smoller, Jordan W.; Stefánsson, Kári; Heel, David A. van; Walters, Robin; Zöllner, Sebastian; Americas, Biobank of the; Project, None Biobank Japan; BioMe,; BioVU,; Study, CanPath - Ontario Health; Group, China Kadoorie Biobank Collaborative; Medicine, Colorado Center for Personalized; Genetics, deCODE; Biobank, FinnGen Estonian; Scotland, Generation; Team, None Genes & Health Research; LifeLines, None; Biobank, Mass General Brigham; Initiative, None Michigan Genomics; Korea, National Biobank of; BioBank, Penn Medicine; Biobank, Qatar; Study, The Qskin Sun and Health; Biobank, Taiwan; Study, The Hunt; Initiative, Ucla Atlas Community Health; Resource, Uganda Genome; Biobank, UK; Martin, Alicia R.; Willer, Cristen J.; Daly, Mark J.; Neale, Benjamin M.

doi:10.1016/j.xgen.2022.100210

Cited by 32 publications

(15 citation statements)

References 85 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To identify putative causal variants in associated loci, we conducted statistical fine-mapping of mtDNA traits in UKB using cross-ancestry meta-analysis summary statistics. While we previously showed that fine-mapping a meta-analysis is often miscalibrated due to heterogeneous characteristics of constituent cohorts (e.g., genotyping or imputation) (Kanai et al, 2022), a within-cohort cross-ancestry meta-analysis like the present study is a notable exception given no such heterogeneity systematically exists across ancestries. We used FINEMAP-inf and SuSiE-inf which model infinitesimal effects (Cui et al, 2022), with cross-ancestry meta-analysis summary statistics (Methods) and a covariate-adjusted in-sample dosage LD matrix (Kanai et al, 2021).…”

Section: Fine-mapping In Ukbmentioning

confidence: 59%

Nuclear genetic control of mtDNA copy number and heteroplasmy in humans

Gupta

Kanai

Durham

et al. 2023

Preprint

Self Cite

View full text Add to dashboard Cite

Human mitochondria contain a high copy number, maternally transmitted genome (mtDNA) that encodes 13 proteins required for oxidative phosphorylation. Heteroplasmy arises when multiple mtDNA variants co-exist in an individual and can exhibit complex dynamics in disease and in aging. As all proteins involved in mtDNA replication and maintenance are nuclear-encoded, heteroplasmy levels can, in principle, be under nuclear genetic control, however this has never been shown in humans. Here, we develop algorithms to quantify mtDNA copy number (mtCN) and heteroplasmy levels using blood-derived whole genome sequences from 274,832 individuals of diverse ancestry and perform GWAS to identify nuclear loci controlling these traits. After careful correction for blood cell composition, we observe that mtCN declines linearly with age and is associated with 92 independent nuclear genetic loci. We find that nearly every individual carries heteroplasmic variants that obey two key patterns: (1) heteroplasmic single nucleotide variants are somatic mutations that accumulate sharply after age 70, while (2) heteroplasmic indels are maternally transmitted as mtDNA mixtures with resulting levels influenced by 42 independent nuclear loci involved in mtDNA replication, maintenance, and novel pathways. These nuclear loci do not appear to act by mtDNA mutagenesis, but rather, likely act by conferring a replicative advantage to specific mtDNA molecules. As an illustrative example, the most common heteroplasmy we identify is a length variant carried by >50% of humans at position m.302 within a G-quadruplex known to serve as a replication switch. We find that this heteroplasmic variant exerts cis-acting genetic control over mtDNA abundance and is itself under trans-acting genetic control of nuclear loci encoding protein components of this regulatory switch. Our study showcases how nuclear haplotype can privilege the replication of specific mtDNA molecules to shape mtCN and heteroplasmy dynamics in the human population.

show abstract

Section: Fine-mapping In Ukbmentioning

confidence: 59%

Nuclear genetic control of mtDNA copy number and heteroplasmy in humans

Gupta

Kanai

Durham

et al. 2023

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…Second, as we used external LD reference panels for PRS construction, PRS performance decreases with LD mismatch between the discovery population and LD reference panel, especially using multi-ancestry GWAS. While we show that LD reference panel differences have a relatively modest effect on PRS accuracy, they have a much larger effect on finemapping 42 , so future efforts are warranted to share in-sample LD without direct access to individual-level genotypes, especially for large consortia with numerous and diverse cohorts. Alternatively, developing more sophisticated individual-level PRS methods that preserve privacy and are scalable to current biobank-scale genomics data is also promising.…”

Section: Limitations Of the Studymentioning

confidence: 71%

Polygenic prediction across populations is influenced by ancestry, genetic architecture, and methodology

Wang

Kanai

Tan

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

Polygenic risk scores built from multi-ancestry genome-wide association studies (GWAS, PRSmulti) have the potential to improve PRS accuracy and generalizability across populations. To provide the best practice to leverage the increasing diversity of genomic studies, we used large-scale simulated and empirical data to investigate how ancestry composition, trait-specific genetic architecture, and PRS methodology affect the performance of PRSmulti as compared to PRS constructed from single-ancestry GWAS (PRSsingle). In both simulations on 6 various scenarios and empirical analyses on 17 anthropometric and blood panel traits, we showed that the accuracy of PRSmulti overall outperformed PRSsingle in the understudied target populations, except for a few comparisons where the understudied population only accounted for a very small proportion of the multi-ancestry GWAS. Further, using substantially fewer samples for traits such as height and mean corpuscular volume from Biobank Japan (BBJ) may achieve comparable accuracies to using 320,000 European (EUR) individuals from UK Biobank (UKBB). Finally, we find that incorporating PRS based on local ancestry-informed GWAS and large-scale EUR-based PRS improved predictive performance than using EUR-based PRS alone in understudied African (AFR) population, especially for less polygenic traits when there are variants with large ancestry-specific effects. Overall, our study provides insights into how ancestry composition and genetic architecture impact polygenic prediction across populations, particularly across imbalanced sample sizes. Our work also highlights the need for increasing diversity in genetic studies to achieve equitable PRS performance across ancestral populations and provides practical guidance on developing PRS from multiple resources.

show abstract

“…To test if the recall of the MR methods could be increased by meta-analysing eQTLs across multiple studies from the same tissue, we obtained the cis -eQTL summary statistics from the AdipoExpress project, a meta-analysis of five subcutaneous adipose tissue studies (n = 2,344). Although AdipoExpress was not able to use SuSiE for fine-mapping due to the risk of false positives [21], they used all-but-one conditional analysis [22] to identify conditionally distinct signals. After converting these conditional summary statistics to approximate Bayes factors (see Methods), we performed colocalisation with INTERVAL pQTLs using the same workflow that we previously used for the eQTL Catalogue datasets.…”

Section: Resultsmentioning

confidence: 99%

“…Since AdipoExpess analysis used the GRCh37 reference genome, we first converted the variant positions to GRCh38 coordinates with the MungeSummstats [62] R package. Since using SuSiE to identify conditionally distinct signals is prone to false positives in a meta-analysis setting [21], AdipoExpress used all-but-one conditional meta-analysis and also released summary statistics for distinct signals conditioned on all other significant signals in the same cis- region. To use these results with coloc.susie, we first converted the all-but-one conditional betas and standard errors to log approximate Bayes factors (LABFs) using the process.dataset() function from the coloc R package.…”

Section: Methodsmentioning

confidence: 99%

Extensive co-regulation of neighbouring genes complicates the use of eQTLs in target gene prioritisation

Tambets,

Kolde,

Kolberg

et al. 2023

Preprint

View full text Add to dashboard Cite

Identifying causal genes underlying genome-wide association studies (GWAS) is a fundamental problem in human genetics. Although colocalisation with gene expression quantitative trait loci (eQTLs) is often used to prioritise GWAS target genes, systematic benchmarking has been limited due to unavailability of large ground truth datasets. Here, we re-analysed plasma protein QTL data from 3,301 individuals of the INTERVAL cohort together with 131 eQTL Catalogue datasets. Focusing on variants located within or close to the affected protein identified 793 proteins with at least onecis-pQTL where we could assume that the most likely causal gene was the gene coding for the protein. We then benchmarked the ability ofcis-eQTLs to recover these causal genes by comparing three Bayesian colocalisation methods (coloc.susie, coloc.abf and CLPP) and five Mendelian randomisation (MR) approaches (three varieties of inversevariance weighted MR, MR-RAPS, and MRLocus). We found that assigning fine-mapped pQTLs to their closest protein coding genes outperformed all colocalisation methods regarding both precision (71.9%) and recall (76.9%). Furthermore, the colocalisation method with the highest recall (coloc.susie - 46.3%) also had the lowest precision (45.1%). Combining evidence from multiple conditionally distinct colocalising QTLs with MR increased precision to 81%, but this was accompanied by a large reduction in recall to 7.1%. Furthermore, the choice of the MR method greatly affected performance, with the standard inverse-variance weighted MR often producing many false positives. Our results highlight that linking GWAS variants to target genes remains challenging with eQTL evidence alone, and prioritising novel targets requires triangulation of evidence from multiple sources.

show abstract

Meta-analysis fine-mapping is often miscalibrated at single-variant resolution

Cited by 32 publications

References 85 publications

Nuclear genetic control of mtDNA copy number and heteroplasmy in humans

Nuclear genetic control of mtDNA copy number and heteroplasmy in humans

Polygenic prediction across populations is influenced by ancestry, genetic architecture, and methodology

Extensive co-regulation of neighbouring genes complicates the use of eQTLs in target gene prioritisation

Contact Info

Product

Resources

About