Extended Data Fig. 1: Variance-covariance matrices of observed and imputed genotypes.
From: Mendelian imputation of parental genotypes improves estimates of direct genetic effects
Each matrix shows the variance-covariance matrix within the nuclear family given observed and imputed genotypes. The labels a) to o) correspond to those in Fig. 1. gi1 denotes the proband’s genotype, gi2 denotes the proband’s sibling’s genotype, gp(i) denotes father’s genotype and gm(i) denotes mother’s genotype. A caret (‘hat’) over the ‘g’, such as \(\hat g_{p(i)}\), indicates an unobserved genotype that is imputed (either linearly or non-linearly) using observed genotypes. Displayed are the variance-covariance matrices of the genotypes, normalized by the variance of an observed genotype, 2f(1−f) where f is the allele frequency. This scaling means that the diagonal entry corresponding to an observed genotype is 1. Best linear unbiased imputations can be derived from applying formulae for multivariate Gaussian random-variables to a) (Supplementary Note Appendix D).
