Global Individual Ancestry Using Principal Components for Family Data

Abstract

Studies of complex human diseases and traits associated with candidate genes are potentially vulnerable to bias (confounding) due to population stratification and inbreeding, especially in admixed population. In GWAS, the principal components (PCs) method provides a global ancestry value per subject, allowing corrections for population stratification. However, these coefficients are typically estimated assuming unrelated individuals, and if family structure is present and ignored, such substructures may induce artifactual PCs. Extensions of the PCs method have been proposed by Konishi and Rao [Biometrika 1992;79:631-641], taking into account only siblings' relatedness, and by Oualkacha et al. [Stat Appl Genet Mol Biol 2012, DOI: 10.2202/1544-6115.1711], taking into account large pedigrees and high-dimensional phenotype data. In this work, we extend these methods to estimate the global individual ancestry coefficients from PCs derived from different variance component matrix estimators using SNPs from two simulated data sets and two real data sets: the GENOA sibship data consisting of European and African-American subjects and the Baependi Heart Study consisting of 80 extended Brazilian families, both with genotyping data from the Affymetrix 6.0 chip. Our results show that the family structure plays an important role in the estimation of the global individual ancestry value for extended pedigrees but not for sibships.

Cohort profile: the Baependi Heart Study—a family-based, highly admixed cohort study in a rural Brazilian town

“…Therefore researchers used simulations from two datasets including the Baependi study to derive coefficients which incorporate the relatedness of individuals within family-based designs. 31 These estimates suggest that family structure is important for the estimation of global individual ancestry for extended pedigrees, but not for siblings. A second key statistical output from the cohort concerns the complexity of multiple CVD phenotypes, which makes disease diagnosis and genetic dissection difficult.…”

Identification of causality in genetics and neuroscience

“…In the regression analyses, we considered the main effects for sex, age, and squared age, as well as, the interaction effect between sex and age. To correct for population stratification, we also considered as covariates two measurements of the Baependi population ancestry derived from the two first principal components of heritability for 8,764 single nucleotide polymorphisms (SNPs) (De Andrade et al, 2015;Oualkacha et al, 2012).…”

“…In Genetics, besides understanding how the variables are related to each other, it is crucial to unravel how the connections among the variables are influenced by genetic and environmental factors. Family-based studies have proven to be useful in elucidating genetic and environmental factors affecting variables, since familial clustering may reveal genetic effects (Almasy and Blangero, 1998;Amos, 1994;De Andrade et al, 2015;Lange, 2003;Oualkacha et al, 2012). However, few existing approaches are able to address structure learning of PGMs and family data analysis jointly.…”

“…Thus, (co)variance of phenotypes may be partitioned according to genetic and environmen-INTRODUCTION 7 tal influences, elucidating the nature of their mechanisms. For a multivariate analysis of multiple traits, the multivariate polygenic mixed model can be used (De Andrade et al, 2015;Oualkacha et al, 2012).…”

Herdabilidade e Estudo de Associação Genômica Ampla (GWAS) de atividade física autorreportada: uma investigação de famílias brasileiras

“…ASSOCIAÇÃO DE SNPS E CNVS COM FENÓTIPOS DA ATIVIDADE FÍSICA 21 vidade física, AFT, WK, MPA, VPA, e sedentarismo, utilizamos ainda derivações mais complexas desse modelo poligênico, as quais incluiram, além dos SNPs e CNVs, idade, sexo e as demais covariáveis selecionadas nos ajustes de modelo específicos a cada fenótipo. Ainda, incluiu-se coeficientes de ancestralidade (pca1 e pca2) obtidos pela análise de componentes principais em[De Andrade et al, 2015], os quais funcionam como uma medida de correção importante em estudos de associação com estrutura familiar. Por conta da normalidade das variáveis resposta ser uma das premissas do modelo linear misto poligênico, devendo ser atendida para uma estimação fidedigna dos coeficientes β e p-valores [Almasy eBlangero, 2010], utilizamos transformações dessas variáveis de acordo com os resultados dos ajustes dos modelos, as quais estão descritas no capítulo 4.3.…”