A bootstrap approach to estimating power for linkage heterogeneity

Area of Science:

• Genetics
• Statistical Genetics
• Bioinformatics

Background:

• Linkage heterogeneity occurs when families in a study are linked to different genetic loci.
• Detecting linkage heterogeneity is crucial for accurate genetic mapping and understanding complex diseases.
• Existing methods often assume homogeneity, potentially masking true genetic architecture.

Purpose of the Study:

• To evaluate the statistical power of detecting linkage heterogeneity under various genetic models.
• To determine the optimal conditions for identifying families linked to distinct loci or unlinked to any locus.
• To assess the performance of a bootstrap approach for significance estimation.

Main Methods:

• Simulated pedigrees under null (single locus linkage) and alternative (multiple loci or no linkage) hypotheses.
• Bootstrap resampling to estimate significance levels and power.
• Likelihood ratio tests to assess the strength of evidence for linkage heterogeneity.
• Analysis of the Duke Familial Alzheimer Disease dataset.

Main Results:

• Detection power is highest with zero recombination fraction and equal family proportions linked to two loci (A and B).
• Power diminishes as recombination fraction increases, the proportion of unlinked families rises, or the ratio of linked families becomes unequal.
• For the Alzheimer Disease dataset, power reached 0.94 with a 10:1 likelihood ratio criterion, corresponding to a p-value of 0.013.

Conclusions:

• The bootstrap approach effectively estimates the power to detect linkage heterogeneity.
• Optimal conditions for detecting heterogeneity involve minimal recombination and balanced family distribution across loci.
• Findings provide a framework for designing and interpreting genetic linkage studies, particularly for complex diseases like Alzheimer's.