Unsupervised Ensemble Learning for Efficient Integration of Pre-trained Polygenic Risk Scores DOI Open Access
Chenyin Gao, Justin D. Tubbs, Yi Han

et al.

medRxiv (Cold Spring Harbor Laboratory), Journal Year: 2025, Volume and Issue: unknown

Published: Jan. 6, 2025

The growing availability of pre-trained polygenic risk score (PRS) models has enabled their integration into real-world applications, reducing the need for extensive data labeling, training, and calibration. However, selecting most suitable PRS model a specific target population remains challenging, due to issues such as limited transferability, het-erogeneity, scarcity observed phenotype in settings. Ensemble learning offers promising avenue enhance predictive accuracy genetic assessments, but existing methods often rely on or additional genome-wide association studies (GWAS) from optimize ensemble weights, limiting utility real-time implementation. Here, we present UN supervised en Semble ( UNSemblePRS ), an unsupervised framework, that combines without requiring summaries population. Unlike traditional approaches, aggregates based prediction concordance across curated subset candidate models. We evaluated using both continuous binary traits All Us database, demonstrating its scalability robust performance diverse populations. These results underscore accessible tool integrating contexts, offering broad applicability continues expand.

Language: Английский

Polygenic scoring accuracy varies across the genetic ancestry continuum DOI Creative Commons
Yi Ding, Kangcheng Hou,

Ziqi Xu

et al.

Nature, Journal Year: 2023, Volume and Issue: 618(7966), P. 774 - 781

Published: May 17, 2023

Abstract Polygenic scores (PGSs) have limited portability across different groupings of individuals (for example, by genetic ancestries and/or social determinants health), preventing their equitable use 1–3 . PGS has typically been assessed using a single aggregate population-level statistic R 2 ) 4 , ignoring inter-individual variation within the population. Here, large and diverse Los Angeles biobank 5 (ATLAS, n = 36,778) along with UK Biobank 6 (UKBB, 487,409), we show that accuracy decreases individual-to-individual continuum 7 in all considered populations, even traditionally labelled ‘homogeneous’ ancestries. The decreasing trend is well captured continuous measure distance (GD) from training data: Pearson correlation −0.95 between GD averaged 84 traits. When applying models trained on as white British UKBB to European ATLAS, furthest decile 14% lower relative closest decile; notably, Hispanic Latino American similar performance significantly correlated estimates themselves for 82 traits, further emphasizing importance incorporating interpretation. Our results highlight need move away discrete ancestry clusters towards when considering PGSs.

Language: Английский

Citations

156
Principles and methods for transferring polygenic risk scores across global populations DOI
Linda Kachuri, Nilanjan Chatterjee, Jibril Hirbo

et al.

Nature Reviews Genetics, Journal Year: 2023, Volume and Issue: 25(1), P. 8 - 25

Published: Aug. 24, 2023

Language: Английский

Citations

132
Global Biobank analyses provide lessons for developing polygenic risk scores across diverse cohorts DOI Creative Commons
Ying Wang, Shinichi Namba, Esteban A. Lopera-Maya

et al.

Cell Genomics, Journal Year: 2023, Volume and Issue: 3(1), P. 100241 - 100241

Published: Jan. 1, 2023

Polygenic risk scores (PRSs) have been widely explored in precision medicine. However, few studies thoroughly investigated their best practices global populations across different diseases. We here utilized data from Global Biobank Meta-analysis Initiative (GBMI) to explore methodological considerations and PRS performance 9 biobanks for 14 disease endpoints. Specifically, we constructed PRSs using pruning thresholding (P + T) PRS-continuous shrinkage (CS). For both methods, a European-based linkage disequilibrium (LD) reference panel resulted comparable or higher prediction accuracy compared with several other non-European-based panels. PRS-CS overall outperformed the classic P T method, especially endpoints SNP-based heritability. Notably, is heterogeneous endpoints, biobanks, ancestries, asthma, which has known variation prevalence populations. Overall, provide lessons construction, evaluation, interpretation GBMI resources highlight importance of biobank-scale genomics era.

Language: Английский

Citations

69
Recent advances in polygenic scores: translation, equitability, methods and FAIR tools DOI Creative Commons
Ruidong Xiang, Martin Kelemen, Yu Xu

et al.

Genome Medicine, Journal Year: 2024, Volume and Issue: 16(1)

Published: Feb. 19, 2024

Abstract Polygenic scores (PGS) can be used for risk stratification by quantifying individuals’ genetic predisposition to disease, and many potentially clinically useful applications have been proposed. Here, we review the latest potential benefits of PGS in clinic challenges implementation. could augment through combined use with traditional factors (demographics, disease-specific factors, family history, etc.), support diagnostic pathways, predict groups therapeutic benefits, increase efficiency clinical trials. However, there exist maximizing utility PGS, including FAIR (Findable, Accessible, Interoperable, Reusable) standardized sharing genomic data needed develop recalculate equitable performance across populations ancestries, generation robust reproducible calculations, responsible communication interpretation results. We outline how these may overcome analytically more diverse as well highlight sustained community efforts achieve equitable, impactful, healthcare.

Language: Английский

Citations

29
Integrative polygenic risk score improves the prediction accuracy of complex traits and diseases DOI Creative Commons
Buu Truong, Leland E. Hull, Yunfeng Ruan

et al.

Cell Genomics, Journal Year: 2024, Volume and Issue: 4(4), P. 100523 - 100523

Published: March 19, 2024

Polygenic risk scores (PRSs) are an emerging tool to predict the clinical phenotypes and outcomes of individuals. We propose PRSmix, a framework that leverages PRS corpus target trait improve prediction accuracy, PRSmix+, which incorporates genetically correlated traits better capture human genetic architecture for 47 32 diseases/traits in European South Asian ancestries, respectively. PRSmix demonstrated mean accuracy improvement 1.20-fold (95% confidence interval [CI], [1.10; 1.3]; p = 9.17 × 10

Language: Английский

Citations

28
Wrestling with Social and Behavioral Genomics: Risks, Potential Benefits, and Ethical Responsibility DOI Open Access
Michelle N. Meyer, Paul S. Appelbaum, Daniel J. Benjamin

et al.

The Hastings Center Report, Journal Year: 2023, Volume and Issue: 53(S1)

Published: March 1, 2023

Abstract In this consensus report by a diverse group of academics who conduct and/or are concerned about social and behavioral genomics (SBG) research, the authors recount often‐ugly history scientific attempts to understand genetic contributions human behaviors outcomes. They then describe what current science—including genomewide association studies polygenic indexes—can cannot tell us, as well its risks potential benefits. conclude with discussion responsible behavior in context SBG research. research that compares individuals within according “sensitive” phenotype requires extra attention communication findings. (1) on sensitive phenotypes (2) two or more groups defined (a) race, (b) ethnicity, (c) ancestry (where could easily be misunderstood race ethnicity) compelling justification conducted, funded, published. All agree at least convincing argument study's design yield scientifically valid results; some would additionally require study have socially favorable risk‐benefit profile .

Language: Английский

Citations

41
More than a decade of genetic research on the Denisovans DOI
Stéphane Peyrégne, Viviane Slon, Janet Kelso

et al.

Nature Reviews Genetics, Journal Year: 2023, Volume and Issue: 25(2), P. 83 - 103

Published: Sept. 18, 2023

Language: Английский

Citations

27
Aspiring toward equitable benefits from genomic advances to individuals of ancestrally diverse backgrounds DOI Creative Commons
Ying Wang, Yixuan He,

Yue Shi

et al.

The American Journal of Human Genetics, Journal Year: 2024, Volume and Issue: 111(5), P. 809 - 824

Published: April 19, 2024

Language: Английский

Citations

11
Screening embryos for polygenic disease risk: a review of epidemiological, clinical, and ethical considerations DOI Creative Commons
Antonio Capalbo,

G. de Wert,

Heidi Mertes

et al.

Human Reproduction Update, Journal Year: 2024, Volume and Issue: 30(5), P. 529 - 557

Published: May 28, 2024

The genetic composition of embryos generated by in vitro fertilization (IVF) can be examined with preimplantation testing (PGT). Until recently, PGT was limited to detecting single-gene, high-risk pathogenic variants, large structural and aneuploidy. Recent advances have made genome-wide genotyping IVF feasible affordable, raising the possibility screening for their risk polygenic diseases such as breast cancer, hypertension, diabetes, or schizophrenia. Despite a heated debate around this new technology, called embryo (PES; also PGT-P), it is already available patients some countries. Several articles studied epidemiological, clinical, ethical perspectives on PES; however, comprehensive, principled review emerging field missing.

Language: Английский

Citations

11
Strategies for the Genomic Analysis of Admixed Populations DOI Creative Commons

Taotao Tan,

Elizabeth G. Atkinson

Annual Review of Biomedical Data Science, Journal Year: 2023, Volume and Issue: 6(1), P. 105 - 127

Published: April 26, 2023

Admixed populations constitute a large portion of global human genetic diversity, yet they are often left out genomics analyses. This exclusion is problematic, as it leads to disparities in the understanding structure and history diverse cohorts performance genomic medicine across populations. have particular statistical challenges, inherit segments from multiple source populations-the primary reason historically been excluded studies. In recent years, however, an increasing number methods software tools developed account for leverage admixture context Here, we provide survey such computational strategies informed consideration allow well-calibrated inclusion mixed ancestry large-scale studies, detail persisting gaps existing tools.

Language: Английский

Citations

21