UK Biobank | Tanigawa Lab

PRISM: ancestry-aware integration of tissue-specific genomic annotations enhances the transferability of polygenic scores

Thu, 13 Nov 2025 00:00:00 +0000

Accurate genetic prediction of complex traits has the potential to substantially reduce the disease burden, but the limited transferability of polygenic score (PGS) across genetic ancestry groups remains a major challenge.

Here we propose to integrate three major approaches to address the PGS transferability. We develop PRISM and apply it to 7352 fine-mapped variants from MVP, 414 ENCODE annotations, and 406,659 individuals from the UK Biobank.

Using PRISM, we evaluated the effects of ancestry and tissue-specific integration of genomic annotations. We also compared our approach with existing methods and investigated which genomic annotation would be the most informative.

We found that tissue-matched genomic annotations (from ENCODE) are most effective for enhancing PGS transferability, even though the tissue-agnostic strategy can leverage a ~18-fold larger number of genomic annotation datasets.

Similarly, the use of genetic ancestry-specific fine-mapped variants is most effective, despite the power difference. Our approach offers a pragmatic solution for working with data with uneven coverage across contexts.

Our benchmarking analysis shows PRISM benefits from three different strategies for enhancing PGS transferability.

We showed, through feature importance analysis, that having one reference genomic annotation is not sufficient, highlighting the advantage of systematic integration with our approach.

We confirm that the genetic variants selected from our approach are supported by various genomic annotations compared to other variants in LD.

I am grateful to Lucy and the team for making this work possible.

Hypometric genetics: Improved power in genetic discovery by incorporating quality control flags

Tue, 22 Oct 2024 00:00:00 +0000

We introduce “hypometric genetics,” an approach to investigate the genetic basis of binarized traits representing the presence of below-the-limit-of-quantification (BLQ) quality control indicators.

A polygenic score method boosted by non-additive models

Wed, 29 May 2024 00:00:00 +0000

We developed GenoBoost, a polygenic score modeling approach, incorporating both additive and non-additive genetic dominance effects.

Power of inclusion: Enhancing polygenic prediction with admixed individuals

Thu, 26 Oct 2023 00:00:00 +0000

Polygenic score (PGS), a statistical approach to estimating genetic predisposition on traits, attracted substantial research interest. The current PGS models show limited transferability across populations, and there are a number of great new methods to address this challenge. We propose inclusive polygenic score (iPGS), a PGS training strategy to capture ancestry-shared genetic effects by analyzing individuals across the continuum of genetic ancestry. We work directly on the individual-level data without relying on GWAS results and LD references.

We tested our approach across 33 simulation configurations and 60 quantitative traits in UK Biobank. We see increased power by including ancestry-diverse individuals compared to our baseline model trained only on white British individuals.

We observe improvements in performance for all population groups. The average improvement across the 60 traits was 60.8% for African, 11.6% for South Asian, 7.3% for non-British White, 4.8% for White British, and 17.8% for other diverse individuals.

To consider ancestry-dependent genetic effects on top of ancestry-shared effects, we developed iPGS+refit. We used a heterogeneity test in GWAS meta-analysis and identified genetic variants with heterogeneous associations, such as the ACKR1 locus for neutrophil count.

Our iPGS+refit starts with one ancestry-shared component (iPGS) and adds ancestry-dependent effects using a relatively small number of genetic loci, facilitating better interpretation. We used hematological traits to show improved predictive performance.

We compared our model with PRS-CSx, a commonly-used multi-ancestry PGS method from summary statistics from multiple population groups and ancestry-matched reference panels. In our analysis, our iPGS/iPGS+refit models showed competitive or improved performance.

We thank UK Biobank, its participants, amazing collaborators and colleagues, as well as funding.

You can browse and download our iPGS models at our iPGS browser. Taking advantage of the sparsity of our PGS models, it offers direct integration with HaploReg and GREAT.

Large-scale multivariate sparse regression with applications to UK Biobank

Tue, 19 Jul 2022 00:00:00 +0000

In this study led by Junyang Qian, we present a method to fit sparse multi-variate and multi-response regression model.

Significant Sparse Polygenic Risk Scores across 813 traits in UK Biobank

Thu, 24 Mar 2022 00:00:00 +0000

We performed a systematic assessment of the predictive performance of PRS models across >1,500 traits in UK Biobank and report 813 PRS models with significant predictive performance.

Bayesian model comparison for rare-variant association studies of multiple phenotypes

Thu, 25 Nov 2021 00:00:00 +0000

When applied to cardiometabolic biomarker traits in UK Biobank, MRP identified gene-biomarker associations that were not identified in single-variant GWAS analysis.

A cross-population atlas of genetic associations for 220 human phenotypes

Thu, 30 Sep 2021 00:00:00 +0000

Using a set of GWAS summary statistics of diseases characterized from both European (UK Biobank and FinnGen) and East Asian (Biobank Japan) populations, we dissected latent DeGAs components of multi-ethnic association summary statistics.

Genetics of 35 blood and urine biomarkers in the UK Biobank

Wed, 07 Apr 2021 00:00:00 +0000

Through UK Biobank-wide GWAS meta-analysis, we report > 10,000 GWAS associations (p < 5e-9) across 35 biomarkers and >5,700 loci. This includes > 450 large-effect (>0.1 s.d.) associations on protein-altering variants, which we highlight in a Fuji plot (gene symbols).

With FINEMAP, we identified >27,000 distinct signals in >5,000 regions across the 35 traits, of which >2,500 signals were fine-mapped to a single variant. We also trained sparse polygenic risk score (PRS) models with Lasso L1- penalized regression using the snpnet package.

To investigate the influence of the identified genetic basis on diseases, we performed single-variant PheWAS, PRS-PheWAS, and causal inference. Motivated by those results, we developed multi-PRS by combining a single-trait PRS model of disease with that of 35 biomarkers.

When we compare the predictive performance, multi-PRS showed improvements over single-trait snpnet PRS across many diseases, which we also replicated in FinnGen, suggesting both large case numbers and multi-trait modeling might complementarily contribute to improving the power.

This work was led by Nasa Sinnott-Armstrong, myself, and Dr. Manuel A. Rivas, with many contributions from colleagues. We thank UK Biobank, FinnGen, their participants, amazing collaborators, and colleagues, as well as funding.

Cardiac Imaging of Aortic Valve Area From 34287 UK Biobank Participants Reveals Novel Genetic Associations and Shared Genetic Comorbidity With Multiple Disease Phenotypes

Fri, 30 Oct 2020 00:00:00 +0000

A fast and scalable framework for large-scale and ultrahigh-dimensional sparse regression with application to the UK Biobank

Fri, 23 Oct 2020 00:00:00 +0000

In this project led by Junyang Qian, we developed BASIL, a novel algorithm to fit large-scale L1 penalized (Lasso) regression model using an iterative procedure, and implemented R snpnet package specially designed for genetic data.

Fast Lasso method for large-scale and ultrahigh-dimensional Cox model with applications to UK Biobank

Tue, 29 Sep 2020 00:00:00 +0000

We propose extending the BASIL/snpnet algorithm to fit the L1 penalized Cox proportional hazards model using a large-scale dataset from a genotyped cohort.

Rare protein-altering variants in ANGPTL7 lower intraocular pressure and protect against glaucoma

Tue, 01 Sep 2020 00:00:00 +0000

From the analysis of more than 500,000 individuals in population cohorts, we identified rare protein-altering variants in ANGPTL7 that reduce the risk of glaucoma.

We identified rare protein-altering variants in ANGPTL7 that lower intraocular pressure and provide protection against glaucoma.

One of the alleles reported in the study (220C) is highly (50x +) enriched in the Finnish population, highlighting the power of a founder population with a prior bottlenecking event in genetic discovery.

With the comprehensive health information in the two studied cohorts, we assess the potential impact of the rare variants on a spectrum of human disorders. We did not find any severe medical consequences.

Our results indicate that ANGPTL7 is a safe and effective therapeutic target for glaucoma.

Sex-specific genetic effects across biomarkers

Tue, 01 Sep 2020 00:00:00 +0000

In this study led by Emily Flynn, we discovered a surprising sex-specificity in the genetics of testosterone.

Assessing Digital Phenotyping to Enhance Genetic Studies of Human Diseases

Thu, 07 May 2020 00:00:00 +0000

Large-scale population-based genotyped biobanks with dense phenotypic information provide opportunities for genetic analysis at scale.

Components of genetic associations across 2,138 phenotypes in the UK Biobank highlight adipocyte biology

Tue, 31 Dec 2019 00:00:00 +0000

While many pleiotropic genetic loci have been identified, how they contribute to phenotypes across traits and diseases is unclear. We developed DeGAs to address this issue.

When analyzing the genetics of complex traits, extreme polygenicity and pervasive pleiotropy are challenges in the interpretation and translational application of genetic findings.

To address this challenge, we propose to introduce latent components of genetic associations.

In DeGAs (Decomposition of Genetic Associations), we identify latent components of genetic association by applying truncated singular-value decomposition (TSVD) on a matrix consisting of genome-wide association summary statistics computed for thousands of phenotypes. Using those components and our quantitative scores, we represent the genetics of a disease as a mixture of different components. This provides a more interpretable view of disease genetics.

When applied to 2000+ phenotypes in UK Biobank, we found that a related set of phenotypes and variants are captured in DeGAs latent space. For example, standing and sitting heights are in the same direction, even though we applied DeGAs to association summary statistics.

When we look at the top two DeGAs components for body mass index (BMI), the top one (PC2) is mainly driven by fat-related traits, whereas the second most important one (PC1) is mainly driven by fat-free traits, providing an enhanced interpretation of the genetics of BMI.

To prioritize genes for experiments, we applied DeGAs to a subset of the dataset consisting of protein-truncating variants and identified PDE3B and GPR151 as the top two candidates for obesity. Our siRNA knockdown of Gpr151 showed a dramatic decrease in lipids in adipocytes!

Some extensions of DeGAs

In the Rivas lab, we have several projects that extend the work presented in DeGAs.

: We propose dPRS, a method to enhance the interpretability of polygenic risk score (PRS) using DeGAs latent components.
: In DeGAs, we took the summary statistics from univariate association scan across genetic variants and phenotypes. We propose a method to directly fit multi-response sparse regression models.

We provide a resource for the research community. We developed as a part of , whose video tutorial is shown above.

The datasets used in the study are available from figshare.

Y. Tanigawa, and M. A. Rivas, Decomposed matrices used for the analysis described in ‘Components of genetic associations across 2,138 phenotypes in the UK Biobank highlight adipocyte biology’. (2019).

Significant shared heritability underlies suicide attempt and clinically predicted probability of attempting suicide

Fri, 04 Jan 2019 00:00:00 +0000

Using two independent datasets from genotyped cohorts (UK Biobank and electronic medical record (EMR) in Vanderbilt University Medical Center), we quantified the heritability estimates of suicide attempts.

Global Biobank Engine: enabling genotype-phenotype browsing for biobank summary statistics

Wed, 05 Dec 2018 00:00:00 +0000

We present Global Biobank Engine as a platform to visualize genome- and phenome-wide associations and perform statistical inference using those association data.

Medical relevance of protein-truncating variants across 337,205 individuals in the UK Biobank study

Tue, 24 Apr 2018 00:00:00 +0000

Using the UK Biobank population cohort, we investigated the genetic effects of Protein-truncating variants (PTVs) and the clinical impacts.