Some diseases with variable age of onset are due to the presence of predisposing gene mutations. Precise estimation of the age-specific cumulative risk (wich are called penetrance function) for mutation carriers is very important for defining prevention strategies and understanding underlying mechanisms of the diseases. When data are obtained from pedigrees ascertained through affected individuals, an adjustment for ascertainment bias is necessary. If the ascertainment is not tinking into account, the penetrance function estimation can be considerably surestimate. During my PhD, I am focus on developping estimation methods to correct the ascertainment biais. Families acsertainment depend on the disease. So, it is important to develop méthods suited to the circumstance. First, I have considered the Genotype Restricted Likelihood, which estimate the penetrance function without bias whatever the ascertainment bias. I have showed that this method is low efficient when only several individuals are tested. When families are ascertained through at least one affected individual (called proband), ascertainment is easier to model. Il have developed an efficient parametric method to estimate the penetrance function, PEL, for Proband's phenotype Excluded Likelihood. We choose a Weibull model to model the age-dependant cumulative risk function. But if the disease is not modeled by a Weibull function, this method exhibits bad performances. So I also developed an non-parametric method, based on the empirical likelihood theory and relied on the "proposant method" developed by Weinberg, which takes into account familial ascertainment. We called this method IDEAL for Index Discarding EuclIdeAn Likelihood.
