EcoSta 2019: Start Registration
View Submission - EcoSta2019
Title: Assessment of genetic impacts from twin study: A mixture distribution approach Authors:  Zonghui Hu - National Institutes of Health (United States) [presenting]
Abstract: It is challenging to identify features that are genetically determined from those environmentally determined. We approach this by assessing the collective genetic impacts on a feature via the differential correlation in monozygotic twins versus dizygotic twins. Since the underlying order in a twin pair is mostly unclear, data are recorded in a random order, and conventional approaches for correlation coefficients are not valid. To handle the issue of missing order, we model twin data under the framework of mixture bivariate distribution. Taking full advantage of the properties of twin data, we construct and estimate under a combined likelihood function. Despite slow convergence associated with mixture distribution estimation, the combined likelihood induces improved convergency and allows effective statistical inference on the collective genetic impacts. The proposed method is applied to a twin study on immune traits.