New correction algorithms for multiple comparisons in case-control multilocus association studies based on haplotypes...

Area of Science:

• Population genetics
• Statistical genetics
• Bioinformatics

Background:

• Multiple comparison issues complicate genetic association studies examining phenotype-genotype links.
• Bonferroni correction, while common, can be overly conservative due to independence assumptions.

Purpose of the Study:

• Develop novel correction algorithms for testing phenotype-multilocus genotype associations, especially with linked loci.
• Improve the accuracy of type I error rate calculations in multilocus association analyses.

Main Methods:

• Developed an exact algorithm to calculate type I error rates for independency tests, feasible for smaller datasets (up to 50 cases/controls) on a PC cluster.
• Created Markov-chain Monte Carlo (MCMC) based algorithms for asymptotic type I error rate calculation, approximating exact values.

Main Results:

• New algorithms applied to simulated and real data showed significantly lower overall type I error rates (one-third to half) compared to Bonferroni correction for linked loci.
• The MCMC method provided a good approximation to the exact calculation results.

Conclusions:

• The developed algorithms provide more accurate type I error control for multilocus association studies involving linkage disequilibrium.
• These methods are valuable for analyzing case-control and cohort study data in population genetics research.