The admixture maximum likelihood test to test for association between rare variants and disease phenotypes

Area of Science:

• Genetics
• Statistical Genetics
• Bioinformatics

Background:

• Advancements in genotyping and sequencing enable rare variant association studies.
• Single variant testing lacks power for rare variants, necessitating burden tests.
• The rare admixture maximum likelihood test (RAML) is an extension of a previous method for analyzing rare variants.

Purpose of the Study:

• To compare the performance of the RAML method with six other rare variant burden tests.
• To evaluate the power of RAML across various simulation scenarios.

Main Methods:

• Simulation testing was employed to assess the power of RAML and other burden tests.
• Scenarios modeled variations in effect size, direction, and proportion of associated variants.
• The RAML method was compared against six established rare variant association testing methods.

Main Results:

• RAML demonstrated the highest power in scenarios with a small proportion of associated variants.
• SKAT-O showed slightly better performance when a higher proportion of variants were associated.
• Power was generally moderate across most scenarios, highlighting the need for large sample sizes.

Conclusions:

• RAML is a flexible burden test applicable to dichotomous and quantitative traits, accommodating covariates.
• The method makes no prior assumptions on the proportion or effect of associated variants.
• RAML performed well across diverse scenarios, underscoring its utility in rare variant association studies.