Linkage analysis under locus heterogeneity: behaviour of the A-test in complex analyses

Area of Science:

• Genetics
• Statistical genetics
• Genetic epidemiology

Background:

• The admixture test (A-test) is widely used for linkage analysis, particularly when locus heterogeneity is suspected.
• Its application extends to various data types, including pairwise, multipoint, and multi-region analyses.
• However, the conditions influencing its reliability across these diverse scenarios are not fully understood.

Purpose of the Study:

• To analytically evaluate the conditions under which the admixture test (A-test) remains consistent and accurate.
• To identify potential sources of bias in linkage parameter estimates when using the A-test.
• To provide guidance on the appropriate application of the A-test in genetic studies.

Main Methods:

• Analytical evaluations of the admixture test's statistical properties.
• Investigation of the impact of phenotype-linkage probability relationships on A-test consistency.
• Assessment of bias in recombination fraction (theta) and proportion of linked families (alpha) estimates under varying family sizes.

Main Results:

• The A-test demonstrates inconsistency when a relationship exists between phenotype and the probability of linkage.
• Biased estimates for recombination fraction (theta) and proportion of linked families (alpha) can arise if linked family frequencies differ between small and large families.
• Inconsistency is noted if phenotype and age-specific penetrance probabilities are unequal between linked and unlinked families.

Conclusions:

• The admixture test (A-test) requires cautious application, especially when phenotype influences linkage probability or family sizes vary.
• Researchers should verify the equality of phenotype and age-specific penetrance probabilities for linked and unlinked families.
• If family size disparities are unavoidable, the degree of bias must be assessed, and size-specific alpha values should be employed for risk calculations.