Estimating gene penetrance from family data

Area of Science:

• Genetics
• Biostatistics
• Epidemiology

Background:

• Penetrance estimation often assumes independent family phenotypes, which is unrealistic with shared risk factors.
• Shared risk factors induce residual correlation in relatives' phenotypes, even after adjusting for a specific gene's genotype.
• Existing methods to address residual correlation lack systematic performance evaluation.

Purpose of the Study:

• To systematically evaluate methods for estimating penetrance in the presence of residual phenotype correlation.
• To compare the performance of different penetrance estimation methods under various genetic and ascertainment scenarios.

Main Methods:

• Simulated genotype data for rare and common alleles with varying penetrance.
• Generated correlated disease survival times using the Clayton-Oakes copula model.
• Ascertained families via population and clinic designs, comparing method estimates to optimal model-based estimates.

Main Results:

• Penetrance estimates for common, low-risk genotypes were more robust to model misspecification than rare, moderate-risk genotypes.
• Ignoring residual correlation or using only segregating families led to biased penetrance estimates for rare alleles.
• A method assuming genetic heterogeneity performed nearly optimally, even with binary outcomes.

Conclusions:

• Penetrance estimates accommodating residual phenotype correlation outperform those that ignore it.
• Coding censored survival outcomes as binary minimally impacts mean-square error if censoring is not extensive.
• Accurate penetrance estimation necessitates addressing correlated family phenotypes.