Path Diagram Illustrating the Relationship between Admixture, Ancestry, and Phenotype

<p>This figure was created based on the rules of path diagrams outlined in [<a href="http://www.plosgenetics.org/article/info:doi/10.1371/journal.pgen.0020137#pgen-0020137-b065" target="_blank">65</a>] with minor modifications. We wish to explore the association of the putative QTL with a given phenotype. However, as illustrated, this zero-order (i.e., unadjusted) association may be affected by relationships with other factors. The rectangles and ellipses in the path diagram represent observable and latent (unobservable) variables, respectively. The dashed ellipses indicate variables potentially capable of influencing the phenotype. Sources of error from random variation introduced by the meiosis process or measurement error are indicated for observable and unobservable variables. The variable <i>v</i>_i, <i>i</i> = 1,2 denotes the number of alleles inherited from a specific parental population at the i^th QTL (the putative QTL, QTL 1, is observed, whereas QTL 2 is unobserved). Note that for a specific QTL, only two possible values of <i>V</i>_i, <i>i</i> = 1,2 are considered in the model; the third possible value will serve as a reference. <i>P</i>_i, <i>i</i> = 1,2 represents the ancestry of each parent for a sampled individual. The objective is to test for association between the putative QTL and the observed phenotype. Observed association may simply result from unaccounted correlation among the putative QTL, the phenotype, and individual ancestry. The association can be further confounded by the presence of unobserved factors, such as QTL 2. Controlling for parental and individual ancestry would break this confounding pathway. Because ancestry is not directly observable, individual admixture estimates are used as surrogates. These estimates, obtainable through existing software, can be seen as error-contaminated measurements of the true individual ancestry values. Hence, the measurement error problem must be addressed when including these estimates in the model. Hoggart et al. [<a href="http://www.plosgenetics.org/article/info:doi/10.1371/journal.pgen.0020137#pgen-0020137-b007" target="_blank">7</a>] offer a figure similar to the one presented here.