tmax

tmax performs association analysis of tag Single Nucleotide Polymorphisms (SNPs) in targeted genomic regions to detect loci associated with complex disease liability by leveraging linkage disequilibrium (LD) among polymorphisms.

Key Features:

• Association of linked loci: Evaluates association of tag SNPs within targeted genomic regions to identify variants related to complex disease liability.
• Tag SNP selection using LD: Selects tag SNPs based on substantial linkage disequilibrium with other polymorphisms to ensure regional coverage.
• T(B): Computes the maximum of a series of test statistics with multiple-testing correction via the Bonferroni procedure.
• T(P): Computes the maximum of test statistics with multiple-testing correction via permutation of case-control status.
• T(S): Tests the maximum of a smoothed curve fitted to the series of test statistics to capture spatially correlated signals.
• T(R) (Hotelling T(2)): Uses a regression-based Hotelling T(2) approach to evaluate linear combinations of tag SNPs.
• Simulation-based performance assessment: Assesses power and performance through simulations that mimic human population data with realistic LD and allele effects associated with disease liability.
• Factors affecting power: Characterizes effects of SNP correlation structure, density of tag SNPs, and placement of the liability allele on test power.
• Comparative performance findings: Identifies that T(S) is most powerful when many highly correlated SNPs are tested, T(P) is superior with optimized SNP selection (genotyping ~10–20 SNPs per gene), and T(P) and T(R) are comparable when minimizing SNPs per gene.

Scientific Applications:

• Complex disease association studies: Detects loci and evaluates associations within candidate genes or targeted regions for complex traits and diseases.
• Candidate gene/region analysis: Tests tag SNPs to interrogate linkage and association within specific genomic regions.
• SNP selection strategy optimization: Informs design choices for SNP density and selection (including genotyping schemes of ~10–20 SNPs per gene) to maximize power.
• Method comparison and power evaluation: Uses simulation frameworks to compare inferential procedures and quantify power under realistic LD and allele-effect scenarios.

Methodology:

Evaluates tag SNPs using four inferential procedures—T(B) (max statistic with Bonferroni correction), T(P) (max statistic with permutation of case-control status), T(S) (max of a smoothed curve fitted to the series of test statistics), and T(R) (Hotelling T(2) regression of linear combinations)—and assesses performance via simulations mimicking human population LD and allele effects.