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

Coombes, Brandon J.; Ploner, Alexander; Bergen, Sarah E.; Biernacka, Joanna M.

doi:10.1002/gepi.22339

Cited by 72 publications

(48 citation statements)

References 44 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Similar to pseudovalidation approaches, no tuning sample is required when assuming an infinitesimal model. 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 [ 17 , 18 ].…”

Section: Introductionmentioning

confidence: 99%

Evaluation of polygenic prediction methodology within a reference-standardized framework

et al. 2021

View full text Add to dashboard Cite

The 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. Eight polygenic score methods were tested: p-value thresholding and clumping (pT+clump), SBLUP, lassosum, LDpred1, LDpred2, PRScs, DBSLMM and SBayesR, evaluating their performance to predict outcomes in UK Biobank and the Twins Early Development Study (TEDS). Strategies to identify optimal p-value thresholds and shrinkage parameters were compared, including 10-fold cross validation, pseudovalidation and infinitesimal models (with no validation sample), and multi-polygenic score elastic net models. LDpred2, lassosum and PRScs performed strongly using 10-fold cross-validation to identify the most predictive p-value threshold or shrinkage parameter, giving a relative improvement of 16–18% over pT+clump in the correlation between observed and predicted outcome values. Using pseudovalidation, the best methods were PRScs, DBSLMM and SBayesR. PRScs pseudovalidation was only 3% worse 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. Within a reference-standardized framework, the best polygenic prediction was achieved using LDpred2, 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

Section: Introductionmentioning

confidence: 99%

Evaluation of polygenic prediction methodology within a reference-standardized framework

et al. 2021

View full text Add to dashboard Cite

show abstract

“…The PRSs were constructed using PRSice2 [ 42 ] to prune ( r 2 > 0.1 within a 500 kb window) and restrict SNPs to a given p value threshold (p t = 0.0001, 0.001, 0.01, 0.05, 0.1, 0.2, 1), with SNP alleles weighted by their log(HR) estimates. We then performed a principal component analysis (PCA) on the set of PRSs estimated at different p value thresholds and used the first PRS principal component to test for association with the outcome; this PRS-PCA strategy eliminates the multiple testing across PRSs based on different p value thresholds [ 43 ]. The PRSs for TR and THR were tested for association with the respective treatment outcome in each left out dataset using Cox proportional hazards models, and the results from the analyses of the three datasets were meta-analyzed to assess overall PRS prediction of treatment response.…”

Section: Methodsmentioning

confidence: 99%

Genetic contributions to alcohol use disorder treatment outcomes: a genome-wide pharmacogenomics study

Biernacka

Coombes

Batzler

et al. 2021

Neuropsychopharmacol.

Self Cite

View full text Add to dashboard Cite

Naltrexone can aid in reducing alcohol consumption, while acamprosate supports abstinence; however, not all patients with alcohol use disorder (AUD) benefit from these treatments. Here we present the first genome-wide association study of AUD treatment outcomes based on data from the COMBINE and PREDICT studies of acamprosate and naltrexone, and the Mayo Clinic CITA study of acamprosate. Primary analyses focused on treatment outcomes regardless of pharmacological intervention and were followed by drug-stratified analyses to identify treatment-specific pharmacogenomic predictors of acamprosate and naltrexone response. Treatment outcomes were defined as: (1) time until relapse to any drinking (TR) and (2) time until relapse to heavy drinking (THR; ≥ 5 drinks for men, ≥4 drinks for women in a day), during the first 3 months of treatment. Analyses were performed within each dataset, followed by meta-analysis across the studies (N = 1083 European ancestry participants). Single nucleotide polymorphisms (SNPs) in the BRE gene were associated with THR (min p = 1.6E−8) in the entire sample, while two intergenic SNPs were associated with medication-specific outcomes (naltrexone THR: rs12749274, p = 3.9E−8; acamprosate TR: rs77583603, p = 3.1E−9). The top association signal for TR (p = 7.7E−8) and second strongest signal in the THR (p = 6.1E−8) analysis of naltrexone-treated patients maps to PTPRD, a gene previously implicated in addiction phenotypes in human and animal studies. Leave-one-out polygenic risk score analyses showed significant associations with TR (p = 3.7E−4) and THR (p = 2.6E−4). This study provides the first evidence of a polygenic effect on AUD treatment response, and identifies genetic variants associated with potentially medication-specific effects on AUD treatment response.

show abstract

“…Following QC, polygenic risk scores (PRS) were derived using common (>5% minor allele frequency; MAF), well-imputed (INFO>0.8) variants using PLINK version 1.9 [ 35 ], based on large discovery GWAS of primarily European ancestry, with no overlap with the target sample: ADHD (19,099 cases and 34,194 controls) [ 36 ], anxiety disorders (31,977 cases, 82,114 controls) [ 10 ], MDD (59,851 cases and 113,154 controls) [ 11 ], schizophrenia (67,390 cases and 94,015 controls) [ 37 ], ASD (18,382 cases, 27,969 controls) [ 38 ], and bipolar disorder (20,352 cases and 31,358 controls) [ 39 ]. For each discovery GWAS, PRS were calculated using 7 different p-value thresholds and the first principal component based on the correlation matrix for these PRS was extracted and analysed, using the PRS-PCA approach [ 40 ]; see details in S1 Text in S1 File . The PRS were standardised to be z-scores for each analysis.…”

Section: Methodsmentioning

confidence: 99%

Examining sex differences in neurodevelopmental and psychiatric genetic risk in anxiety and depression

et al. 2021

View full text Add to dashboard Cite

Anxiety and depression are common mental health disorders and have a higher prevalence in females. They are modestly heritable, share genetic liability with other psychiatric disorders, and are highly heterogeneous. There is evidence that genetic liability to neurodevelopmental disorders, such as attention deficit hyperactivity disorder (ADHD) is associated with anxiety and depression, particularly in females. We investigated sex differences in family history for neurodevelopmental and psychiatric disorders and neurodevelopmental genetic risk burden (indexed by ADHD polygenic risk scores (PRS) and rare copy number variants; CNVs) in individuals with anxiety and depression, also taking into account age at onset. We used two complementary datasets: 1) participants with a self-reported diagnosis of anxiety or depression (N = 4,178, 65.5% female; mean age = 41.5 years; N = 1,315 with genetic data) from the National Centre for Mental Health (NCMH) cohort and 2) a clinical sample of 13,273 (67.6% female; mean age = 45.2 years) patients with major depressive disorder (MDD) from the Psychiatric Genomics Consortium (PGC). We tested for sex differences in family history of psychiatric problems and presence of rare CNVs (neurodevelopmental and >500kb loci) in NCMH only and for sex differences in ADHD PRS in both datasets. In the NCMH cohort, females were more likely to report family history of neurodevelopmental and psychiatric disorders, but there were no robust sex differences in ADHD PRS or presence of rare CNVs. There was weak evidence of higher ADHD PRS in females compared to males in the PGC MDD sample, particularly in those with an early onset of MDD. These results do not provide strong evidence of sex differences in neurodevelopmental genetic risk burden in adults with anxiety and depression. This indicates that sex may not be a major index of neurodevelopmental genetic heterogeneity, that is captured by ADHD PRS and rare CNV burden, in adults with anxiety and depression.

show abstract

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

Cited by 72 publications

References 44 publications

Evaluation of polygenic prediction methodology within a reference-standardized framework

Evaluation of polygenic prediction methodology within a reference-standardized framework

Genetic contributions to alcohol use disorder treatment outcomes: a genome-wide pharmacogenomics study

Examining sex differences in neurodevelopmental and psychiatric genetic risk in anxiety and depression

Contact Info

Product

Resources

About