Abundant pleiotropy has been observed in many complex traits or disease. When pleiotropy exists, testing rare variants for multiple phenotypes simultaneously is often more powerful than for single phenotype through borrowing additional information from cross-phenotype correlation. Likewise, identifying rare variants associated with repeated phenotypic measurements in longitudinal studies can also have greater statistical power than that in a cross-sectional study. On the other hand, functional rare variants are often enriched in family-based designs. Longitudinal familybased designs therefore provide valuable opportunities to increase statistical power on identifying pleiotropic rare variants associated with multiple phenotypes. In addition, identi fi cation of pleiotropic variants is helpful for elucidating shared pathogenesis of multiple phenotypes. However, statistical tests for pleiotropic rare variants detection in longitudinal family studies remain fairly limited. We extended pedigree-based burden and kernel association tests to longitudinal studies with multiple phenotypes. Generalized estimating equation (GEE) approaches were used to account for the correlations from multiple phenotypes at individual time points as well as the complex correlations between repeated measures of the same phenotype (serial correlations) and between individuals within the same family (familial correlations). Extensive simulation studies were conducted to evaluate performance of the proposed tests under various con fi gurations. The proposed tests were illustrated by a real data example. Both simulation study and data example suggested that incorporating multiple phenotypes can increase statistical power of the proposed tests on rare variant detection. (This study has been supported by grants from Ministry of Science and Technology (MOST105-2314-B-400-017) and National Health Research Institutes in Taiwan (PH-104~106-PP-04).)
Date:
2018-10
Relation:
European Journal of Human Genetics. 2018 Oct;26(Suppl.):782.
