Power loss for multiallelic transmission/disequilibrium test when errors introduced: GAW11 simulated data

Area of Science:

• Genetics
• Statistical genetics
• Bioinformatics

Background:

• Transmission/disequilibrium tests (TDT) are utilized for localizing genes associated with complex diseases using genotype trios.
• Random errors, such as allele changes in genotype data, can impact the accuracy and power of these genetic association studies.

Purpose of the Study:

• To evaluate the effect of random errors in genotype trios on the power of the multiallelic TDT (TDTmhet) to detect linkage.
• To identify the sources of power loss associated with increasing error rates.
• To assess the efficacy of Mendelian inconsistency checks in detecting errors and to provide recommendations for researchers.

Main Methods:

• Simulated genotype data for trios (father, mother, child) were generated with a known disease-associated marker.
• Random errors (allele changes) were introduced at varying rates (1%, 5%, 10%).
• Trios with Mendelian inconsistencies were removed, and the power to detect linkage using TDTmhet was recalculated. Monte Carlo simulations were employed to analyze power loss sources.

Main Results:

• Power to detect linkage decreased significantly with increasing error rates: 8% loss at 1%, 16% at 5%, and 48% at 10%.
• At lower error rates (1%, 5%), power loss was primarily due to reduced sample size. At 10% error, both sample size reduction and error introduction contributed substantially.
• Mendelian inconsistency checks detected only about half of the true errors, with detection rates of 58% (1% error), 60% (5% error), and 62% (10% error).

Conclusions:

• Researchers using TDTmhet for linkage analysis should aim to keep genotype error rates below 5%.
• Mendelian consistency checks alone are insufficient for comprehensive error detection; additional methods are recommended.
• Sample size calculations for TDT studies should account for potential genotype errors to maintain adequate statistical power.