Pedigree analysis package vs. MIXD: fitting the mixed model on a large pedigree

Area of Science:

• Quantitative genetics
• Statistical genetics
• Computational biology

Background:

• Complex segregation analysis is crucial for understanding genetic architectures.
• Accurate parameter estimation is essential for reliable genetic analyses.
• Mixed models are frequently used in genetic analyses.

Purpose of the Study:

• To compare the performance of two analysis methods, Pedigree Analysis Package (PAP) and MIXD, for complex segregation analysis under a mixed model.
• To evaluate the accuracy, bias, and convergence properties of parameter estimates obtained from each method.
• To determine the utility of Monte Carlo Markov chain (MCMC) methods versus approximation methods in genetic analyses.

Main Methods:

• Simulation study using a 232-member pedigree.
• Analysis with Pedigree Analysis Package (PAP) using approximate likelihoods.
• Analysis with MIXD using Monte Carlo Markov chain (MCMC) and Gibbs sampling for likelihood estimation.
• Comparison of parameter estimates including major locus genotype means, gene frequency, environmental variance, and heritability.

Main Results:

• PAP provided unbiased estimates for major locus genotype means and gene frequency but biased estimates for environmental variance and heritability.
• A significant portion of PAP analyses failed to converge.
• MIXD produced unbiased and more accurate parameter estimates than PAP, without convergence issues.
• The performance difference was most pronounced for models with extreme residual additive genetic variance.

Conclusions:

• Monte Carlo Markov chain (MCMC) methods, as implemented in MIXD, offer superior accuracy and reliability for complex segregation analysis compared to approximation methods like PAP.
• MIXD is less sensitive to initial parameter configurations, enhancing its practical utility.
• MCMC methods are recommended when high accuracy in genetic model parameter estimation is required.