Alexandre Bureau

Statistical analysis of cosegregation of rare genetic variants with a disease in families

Family samples are advantageous to detect rare genetic variants involved in complex diseases as such variants are likely to be shared by affected relatives. A variety of statistical tests based on that principle have been developed to link rare variants to the disease under study: RV sharing, GESE and RareIBD. I will describe similarities and differences between these methods, in particular in the way they jointly analyze the rare variants within predefined genomic regions (e.g. genes) to obtain sufficient information, and whether or not they require external estimates of variant frequencies. Results from a simulation study will illustrate how these factors influence power and Type I error of the competing tests. I will also discuss checking the assumption that each variant is introduced only once in a family, which is crucial to the validity of the tests. The RV sharing approach will be illustrated on data from a genomic sequencing study of nonsyndromic oral cleft.