Statistical tests for detecting associations with groups of genetic variants: generalization, evaluation, and...

Area of Science:

• Genetics
• Statistical genetics
• Bioinformatics

Background:

• Genome-Wide Association Studies (GWAS) increasingly focus on rare and uncommon genetic variants.
• Existing methods for rare variant association testing often lack a unified framework.

Purpose of the Study:

• To introduce and evaluate generalized score statistics (GSS) for testing associations between groups of genetic variants and phenotypes.
• To compare the performance of different GSS weighting schemes based on variant characteristics.

Main Methods:

• Described GSS as a weighted sum of single-variant statistics and their cross-products.
• Evaluated power of various weighting schemes against variant minor allele frequency (MAF), proportion associated, and direction of effect.
• Demonstrated that many current rare variant tests are specific instances of GSS.

Main Results:

• Identified two classical statistical tests as robust and powerful within the GSS framework.
• Showcased scenarios where alternative GSS weighting schemes may offer superior performance.
• Provided details on optimal GSS selection based on variant properties.

Conclusions:

• GSS provide a unified and flexible approach to rare variant association testing in GWAS.
• Classical tests remain strong general-purpose tools, but tailored GSS can enhance power in specific situations.
• The CRaVe software package is available for implementing these methods.