Sources of bias for estimators of gene-by-environment interaction from case-parent trio data

The case-parent trio design was popularized by the well-known transmission-disequilibrium test, which conditions on parental genotype information at a test locus to make inference about genetic association that is robust to population stratification. The TDT considers alleles transmitted to the affected offspring (case) as well as those not transmitted (pseudo-controls). The availability of pseudo-controls has lead to a popular belief that the case-parent design itself is robust to population stratification. While this may be true for tests of genetic main effects, some caution is required when extending to inference of gene-by-environment interaction. In this talk, we discuss examples of estimators of gene-by-environment interaction from case-parent trio data and scenarios under which they are biased because of population stratification.

This is joint work with J.-H. Shin and J. Graham.