A principal component approach to improve association testing with polygenic risk scores

Coombes, Brandon J.; Biernacka, Joanna M.

doi:10.1101/847020

Cited by 20 publications

(23 citation statements)

References 26 publications

(35 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We then performed a principal component analysis (PCA) on the resulting PRS and used the first PRS-PCA in subsequent association tests. 40 The PCA reweights the variants included in the PRS to achieve maximum variation across all the p t. 40 This PRS-PCA approach avoids the optimization step in standard pruning and thresholding models to determine the optimal p t , which can inflate type 1 error and result in overfitting. 40 As a sensitivity analysis, we further excluded variants located ±250 kb of the APOE e4 defining SNP, rs429358.…”

Section: Polygenic Risk Scoresmentioning

confidence: 99%

“…40 The PCA reweights the variants included in the PRS to achieve maximum variation across all the p t. 40 This PRS-PCA approach avoids the optimization step in standard pruning and thresholding models to determine the optimal p t , which can inflate type 1 error and result in overfitting. 40 As a sensitivity analysis, we further excluded variants located ±250 kb of the APOE e4 defining SNP, rs429358. The association between each exposure PRS and AD was evaluated using logistic regression adjusting for age, sex, APOE ε4 dose, and 10 principal components.…”

Section: Polygenic Risk Scoresmentioning

confidence: 99%

See 1 more Smart Citation

Causal Associations Between Modifiable Risk Factors and the Alzheimer's Phenome

Andrews

Fulton-Howard

O’Reilly

et al. 2020

Annals of Neurology

View full text Add to dashboard Cite

Objective The purpose of this study was to infer causal relationships between 22 previously reported risk factors for Alzheimer's disease (AD) and the “AD phenome”: AD, AD age of onset (AAOS), hippocampal volume, cortical surface area and thickness, cerebrospinal fluid (CSF) levels of amyloid‐β (Aβ42), tau, and ptau181, and the neuropathological burden of neuritic plaques, neurofibrillary tangles (NFTs), and vascular brain injury (VBI). Methods Polygenic risk scores (PRS) for the 22 risk factors were computed in 26,431 AD cases/controls and the association with AD was evaluated using logistic regression. Two‐sample Mendelian randomization (MR) was used to infer the causal effect of risk factors on the AD phenome. Results PRS for increased education and diastolic blood pressure were associated with reduced risk for AD. MR indicated that only education was causally associated with reduced risk of AD, delayed AAOS, and increased cortical surface area and thickness. Total‐ and LDL‐cholesterol levels were causally associated with increased neuritic plaque burden, although the effects were driven by single nucleotide polymorphisms (SNPs) within the APOE locus. Diastolic blood pressure and pulse pressure are causally associated with increased risk of VBI. Furthermore, total cholesterol was associated with decreased hippocampal volume; smoking initiation with decreased cortical thickness; type 2 diabetes with an earlier AAOS; and sleep duration with increased cortical thickness. Interpretation Our comprehensive examination of the genetic evidence for the causal relationships between previously reported risk factors in AD using PRS and MR supports a causal role for education, blood pressure, cholesterol levels, smoking, and diabetes with the AD phenome. ANN NEUROL 2021;89:54–65

show abstract

Section: Polygenic Risk Scoresmentioning

confidence: 99%

Section: Polygenic Risk Scoresmentioning

confidence: 99%

Causal Associations Between Modifiable Risk Factors and the Alzheimer's Phenome

Andrews

Fulton-Howard

O’Reilly

et al. 2020

Annals of Neurology

View full text Add to dashboard Cite

show abstract

“…All PGS were regressed on genotype batch and 10 principal components to account for possible structural artefacts in the data. Next, in order to have a single polygenic predictor per neurodevelopmental trait in the analyses, we extracted the first principal component from analyses of scores at all 10 thresholds for a given phenotypean approach recently shown by Coombes, Ploner, Bergen, and Biernacka (2020) to maximise prediction while reducing the risk of over-fitting. Full details of the parameters used in generating the PGS and correlations between the PGS at each threshold and the principal component score (henceforth: the PGS) used in these analyses are presented in online Supplementary eMethods 2.…”

Section: Polygenic Scoresmentioning

confidence: 99%

Developmental milestones in early childhood and genetic liability to neurodevelopmental disorders

et al. 2021

View full text Add to dashboard Cite

Background Timing of developmental milestones, such as age at first walking, is associated with later diagnoses of neurodevelopmental disorders. However, its relationship to genetic risk for neurodevelopmental disorders in the general population is unknown. Here, we investigate associations between attainment of early-life language and motor development milestones and genetic liability to autism, attention deficit hyperactivity disorder (ADHD), and schizophrenia. Methods We use data from a genotyped sub-set (N = 25699) of children in the Norwegian Mother, Father and Child Cohort Study (MoBa). We calculate polygenic scores (PGS) for autism, ADHD, and schizophrenia and predict maternal reports of children's age at first walking, first words, and first sentences, motor delays (18 months), and language delays and a generalised measure of concerns about development (3 years). We use linear and probit regression models in a multi-group framework to test for sex differences. Results We found that ADHD PGS were associated with earlier walking age (β = −0.033, padj < 0.001) in both males and females. Additionally, autism PGS were associated with later walking (β = 0.039, padj = 0.006) in females only. No robust associations were observed for schizophrenia PGS or between any neurodevelopmental PGS and measures of language developmental milestone attainment. Conclusions Genetic liabilities for neurodevelopmental disorders show some specific associations with the age at which children first walk unsupported. Associations are small but robust and, in the case of autism PGS, differentiated by sex. These findings suggest that early-life motor developmental milestone attainment is associated with genetic liability to ADHD and autism in the general population.

show abstract

“…Two further methods, SBLUP and SBayesR, can be considered pseudovalidation approaches as they also do not require a tuning sample to identify optimal parameters. Rather than selecting a single tuning parameter, some studies have suggested that combining polygenic scores across p-value thresholds whilst taking into account their correlation using either PCA or model stacking can improve prediction 14,15 .…”

Section: Introductionmentioning

confidence: 99%

Evaluation of Polygenic Prediction Methodology within a Reference-Standardized Framework

Pain

Glanville

Hagenaars

et al. 2020

Preprint

View full text Add to dashboard Cite

BackgroundThe predictive utility of polygenic scores is increasing, and many polygenic scoring methods are available, but it is unclear which method performs best. This study evaluates the predictive utility of polygenic scoring methods within a reference-standardized framework, which uses a common set of variants and reference-based estimates of linkage disequilibrium and allele frequencies to construct scores.MethodsSix polygenic score methods were tested: p-value thresholding and clumping (pT+clump), SBLUP, lassosum, LDPred, PRScs and SBayesR, evaluating their performance to predict outcomes in UK Biobank and the Twins Early Development Study (TEDS). Strategies to identify optimal p-value threshold and shrinkage parameters were compared, including 10-fold cross validation, pseudovalidation (with no validation sample), and multi-polygenic score elastic net models.Resultslassosum, PRScs and LDPred performed strongly using 10-fold cross-validation to identify the most predictive p-value threshold or shrinkage parameter, giving a relative improvement of 14-17% over pT+clump in the correlation between observed and predicted outcome values. Using pseudovalidation, the best method was PRScs, with a relative improvement of >11% over other pseudovalidation methods (lassosum, SBLUP, SBayesR, LDPred), and only 1% less than the best polygenic score identified by 10-fold cross validation. Elastic net models containing polygenic scores based on a range of parameters consistently improved prediction over any single polygenic score.ConclusionWithin a reference-standardized framework, the best polygenic prediction was achieved using lassosum and PRScs, modeling multiple polygenic scores derived using multiple parameters. This study will help researchers performing polygenic score studies to select the most powerful and predictive analysis methods.

show abstract

A principal component approach to improve association testing with polygenic risk scores

Cited by 20 publications

References 26 publications

Causal Associations Between Modifiable Risk Factors and the Alzheimer's Phenome

Causal Associations Between Modifiable Risk Factors and the Alzheimer's Phenome

Developmental milestones in early childhood and genetic liability to neurodevelopmental disorders

Evaluation of Polygenic Prediction Methodology within a Reference-Standardized Framework

Contact Info

Product

Resources

About