Premium
Genetic analysis of biobank data: Familial history aggregation‐based tests (FHAT) with application to Alzheimer's disease
Author(s) -
Wang Yanbing,
Chen Han,
Peloso Gina M.,
Destefano Anita L.,
Dupuis Josée
Publication year - 2020
Publication title -
alzheimer's and dementia
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 6.713
H-Index - 118
eISSN - 1552-5279
pISSN - 1552-5260
DOI - 10.1002/alz.038648
Subject(s) - biobank , association test , disease , proband , type i and type ii errors , family history , genetic association , medicine , genetics , statistics , genotype , biology , gene , mutation , single nucleotide polymorphism , mathematics
Abstract Background Biobanks provide great resources for genetic association analysis. To increase power of genetic association analysis, we propose a novel approach to exploit family history of disease that is often available in biobanks. While methods have been proposed to incorporate family history in single variant analysis, methods to incorporate this information in gene‐based tests, such as the Sequence kernel association test (SKAT), are not available. The goal of this project is to develop a familial history aggregation‐based test (FHAT) to improve power to detect rare variant associations in gene‐based tests, and to apply this method to detect genes associated with Alzheimer’s Disease (AD). Method We assume that participants’ genotypes, and phenotypes of participants’ and their relatives, such as parents, are available from a large cohort. We first assess the association between participants’ genotypes and their phenotypes. Then we evaluate the association between participants’ genotypes and a set of relatives’ disease status conditional on the participants’ disease status. We use a weighted meta‐analysis to combing the score statistics from probands and their relatives. We evaluate the FHAT in simulations, and we apply our novel method to detect associations of gene regions with Alzheimer’s Disease (AD) in the UK Biobank data incorporating parental disease history. Result Our simulation results showed that the type I error of FHAT is well controlled with disease prevalence of 10% and 20%. We obtained corrected type I error rate at various alpha levels using 10,000 simulation replicates with a sample size of 5,000. Out of seven genes previously implicated in AD susceptibility, six showed evidence of association using FHAT in UK Biobank. These six genes had improved significance after incorporating parental phenotype information. Conclusion We proposed a novel approach to include family history, as is often available from medical charts, and showed improved power to detect aggregates of rare variant genetic associations with AD in large cohorts or biobanks.