Genetic Epidemiology最新文献

Polyunsaturated fatty acids (PUFAs) including omega-3 and omega-6 are obtained from diet and can be measured objectively in plasma or red blood cells (RBCs) membrane biomarkers, representing different dietary exposure windows. In vivo conversion of omega-3 and omega-6 PUFAs from short- to long-chain counterparts occurs via a shared metabolic pathway involving fatty acid desaturases and elongase. This analysis leveraged genome-wide association study (GWAS) summary statistics for RBC and plasma PUFAs, along with expression quantitative trait loci (eQTL) to estimate tissue-specific genetically predicted gene expression effects for delta-5 desaturase (FADS1), delta-6 desaturase (FADS2), and elongase (ELOVL2) on changes in RBC and plasma biomarkers. Using colocalization, we identified shared variants associated with both increased gene expression and changes in RBC PUFA levels in relevant PUFA metabolism tissues (i.e., adipose, liver, muscle, and whole blood). We observed differences in RBC versus plasma PUFA levels for genetically predicted increase in FADS1 and FADS2 gene expression, primarily for omega-6 PUFAs linoleic acid (LA) and arachidonic acid (AA). The colocalization analysis identified rs102275 to be significantly associated with a 0.69% increase in total RBC membrane-bound LA levels (p = 5.4 × 10⁻¹²). Future PUFA genetic studies examining long-term PUFA biomarkers are needed to confirm our results.

Genome-wide association studies (GWAS) are hypothesis-free studies that estimate the association between polymorphisms across the genome with a trait of interest. To increase power and to estimate the direct effects of these single-nucleotide polymorphisms (SNPs) on a trait GWAS are often conditioned on a covariate (such as body mass index or smoking status). This adjustment can introduce bias in the estimated effect of the SNP on the trait. Two-sample Mendelian randomisation (MR) studies use summary statistics from GWAS estimate the causal effect of a risk factor (or exposure) on an outcome. Covariate adjustment in GWAS can bias the effect estimates obtained from MR studies conducted using covariate adjusted GWAS data. Multivariable MR (MVMR) is an extension of MR that includes multiple traits as exposures. Here we propose the use of MVMR to correct the bias in MR studies from covariate adjustment. We show how MVMR can recover unbiased estimates of the direct effect of the exposure of interest by including the covariate used to adjust the GWAS within the analysis. We apply this method to estimate the effect of systolic blood pressure on type-2 diabetes and the effect of waist circumference on systolic blood pressure. Our analytical and simulation results show that MVMR provides unbiased effect estimates for the exposure when either the exposure or outcome of interest has been adjusted for a covariate. Our results also highlight the parameters that determine when MR will be biased by GWAS covariate adjustment. The results from the applied analysis mirror these results, with equivalent results seen in the MVMR with and without adjusted GWAS. When GWAS results have been adjusted for a covariate, biasing MR effect estimates, direct effect estimates of an exposure on an outcome can be obtained by including that covariate as an additional exposure in an MVMR estimation. However, the estimated effect of the covariate obtained from the MVMR estimation is biased.

Transcriptome-wide association studies (TWAS) investigate the links between genetically regulated gene expression and complex traits. TWAS involves imputing gene expression using expression quantitative trait loci (eQTL) as predictors and testing the association between the imputed expression and the trait. The effectiveness of TWAS depends on the accuracy of these imputation models, which require genotype and gene expression data from the same samples. However, publicly accessible resources, such as the Genotype Tissue Expression (GTEx) Project, are biased toward individuals of European ancestry, potentially reducing prediction accuracy into other ancestry groups. This study explored eQTL transferability across ancestry groups by comparing two imputation models: PrediXcan (tissue-specific) and UTMOST (cross-tissue). Both models were trained on tissues from the GTEx Project using European ancestry individuals and then tested on data sets of European ancestry and African American individuals. Results showed that both models performed best when the training and testing data sets were from the same ancestry group, with the cross-tissue approach generally outperforming the tissue-specific approach. This study underscores that eQTL detection is influenced by ancestry and tissue context. Developing ancestry-specific reference panels across tissues can improve prediction accuracy, enhancing TWAS analysis and our understanding of the biological processes contributing to complex traits.

Biological datasets often consist of thousands or millions of variables, e.g. genetic variants or biomarkers, and when sample sizes are large it is common to find many associated with an outcome of interest, for example, disease risk in a GWAS, at high levels of statistical significance, but with very small effects. The False Discovery Rate (FDR) is used to identify effects of interest based on ranking variables according to their statistical significance. Here, we develop a complementary measure to the FDR, the priorityFDR, that ranks variables by a combination of effect size and significance, allowing further prioritisation among a set of variables that pass a significance or FDR threshold. Applying to the largest GWAS of type 1 diabetes to date (15,573 cases and 158,408 controls), we identified 26 independent genetic associations, including two newly-reported loci, with qualitatively lower priorityFDRs than the remaining 175 signals. We detected putatively causal type 1 diabetes risk genes using Mendelian Randomisation, and found that these were located disproportionately close to low priorityFDR signals (p = 0.005), as were genes in the IL-2 pathway (p = 0.003). Selecting variables on both effect size and significance can lead to improved prioritisation for mechanistic follow-up studies from genetic and other large biological datasets.

In large cohort studies the number of unaffected individuals outnumbers the number of affected individuals, and the power can be low to detect associations for outcomes with low prevalence. We consider how including recorded family history in regression models increases the power to detect associations between genetic variants and disease risk. We show theoretically and using Monte-Carlo simulations that including a family history of the disease, with a weighting of 0.5 compared with true cases, increases the power to detect associations. This is a powerful approach for detecting variants with moderate effects, but for larger effect sizes a weighting of > 0.5 can be more powerful. We illustrate this both for common variants and for exome sequencing data for over 400,000 individuals in UK Biobank to evaluate the association between the burden of protein-truncating variants in genes and risk for four cancer types.

Gorfine, M., Qu, C.,Peters, U., & Hsu, L. (2024). Unveiling challenges in Mendelian randomization for gene-environment interaction. Genetic Epidemiology, 48, 164–189. https://doi.org/10.1002/gepi.22552