200 MULTIVARIATE POLYGENIC MIXED MODEL IN ADMIXED POPULATION Mariza de Andrade 13 , Júlia Maria Pavan Soler 23 Abstract: In genome-wide association studies (GWAS) the Principal Component based Analysis (PCAs) provides a global ancestry value per subject, allowing corrections for population stratification. These coefficients are typically estimated assuming unrelated individuals and making use of dual-space properties to prevent high dimensional and sparse matrix problems. However, if family structure is present and is ignored, such sub- structure may induce artifactual PCAs. Considering the variable-space in high dimensional data set, extensions of the PCA have been proposed by Konishi and Rao (1992) taking into account only sibship relatedness and by Oualkacha et al. (2012) which can be applied to general pedigrees. Further, considering the subject-space, Blangero et al. (2013) obtained an Eigen simplification of the likelihood function from the univariate polygenic mixed model. In this work we propose to apply these methods to estimate the global individual ancestry using PCs extracted from different variance components matrix estimators and dual-space properties for subjects and variables. We use the GENOA sibship data consisting of European and African American subjects and the Baependi Heart Study consisting of 80 extended families collected from the highly admixture Brazilian population, both with SNPs data from Affymetrix 6.0 chip as applications. All the implementation are done using R package. 1 Introduction Studies of human complex diseases and traits associated with candidate genes are potentially vulnerable to bias (confounding) due to population stratification and inbreeding, especially in admixture population. In genome-wide association studies (GWAS) the Principal Components (PC) method provides a global ancestry value per subject, allowing corrections for population stratification (Price et al., 2006). However, these coefficients are 1 Mayo Clinic - USA 2 IME-USP. e-mail: pavan@ime.usp.br 3 The authors thank Debashree Ray for your contribution to the computational implementation of the methodology as well as the Mayo Clinic (Rochester, MN, USA) and the Laboratory of Genetics and Molecular Cardiology - Heart Institute – University of Sao Paulo (SP, Brazil) for providing the data sets used in this study.