Likelihoods and TDT for the case-parents design

Area of Science:

• Genetics
• Statistical genetics
• Population genetics

Background:

• Family-based association studies, such as the transmission/disequilibrium test (TDT), minimize population stratification bias.
• Existing sample size and power calculations for TDT often assume multiplicative allele effects, potentially leading to inaccuracies.
• Some TDT power calculations overlook allele transmission dependencies in heterozygous parents, particularly when effects are non-multiplicative.

Purpose of the Study:

• To develop accurate sample size and power calculations for the transmission/disequilibrium test (TDT) accommodating general genotype relative risks.
• To present and evaluate likelihood-based methods as alternatives to the TDT for genetic association analyses.
• To compare the efficiency of TDT and likelihood methods under various genetic models.

Main Methods:

• Demonstrated sample size and power calculations for the TDT with general genotype relative risks.
• Developed exact likelihood methods for analyzing genetic association under diverse genotype relative risk models.
• Conducted numerical comparisons to assess the performance of TDT versus likelihood methods.

Main Results:

• The TDT is inefficient for recessive relative risk patterns.
• Assuming multiplicative models for non-rare alleles can significantly underestimate TDT sample size requirements.
• The TDT can be substantially less efficient than likelihood methods for dominant genetic effects with common alleles.
• A two-degree-of-freedom likelihood ratio statistic shows robustness across various genotype relative risk models.

Conclusions:

• Accurate sample size calculations for TDT require considering general genotype relative risks, not just multiplicative models.
• Likelihood methods offer a robust and often more efficient alternative to TDT, particularly under dominant or recessive models.
• Conditional logistic regression can be utilized for assessing allelic effects and gene-environment interactions.