Evaluations of maximization procedures for estimating linkage parameters under heterogeneity

Area of Science:

• Genetics
• Statistical genetics
• Computational biology

Background:

• Locus heterogeneity complicates linkage analysis for complex genetic traits.
• Existing methods often involve maximizing high-dimensional likelihoods.
• Computational challenges arise from multiple parameters in these likelihoods.

Purpose of the Study:

• To address the computational challenges of incorporating locus heterogeneity in genetic analysis.
• To evaluate and compare different likelihood maximization procedures for genetic mapping.
• To demonstrate the application of these methods to real genetic datasets.

Main Methods:

• Exploration of Expectation Maximization (EM) and Stochastic Expectation Maximization (SEM) algorithms.
• General formulation for accounting for heterogeneity.
• Application to specific genetic models and simulation studies.
• Calculation of standard errors (SEs) for parameter estimates.

Main Results:

• SEM demonstrates superior performance compared to EM in simulation studies and real data analysis.
• Limitations of the admixture approach for heterogeneity were illustrated.
• Methods for obtaining confidence intervals using SEs were presented.

Conclusions:

• SEM offers a more effective computational approach for handling locus heterogeneity in genetic linkage analysis.
• The developed methods provide robust parameter estimation and confidence intervals.
• This work advances the accuracy of disease gene mapping for complex traits.