Adaptive combination of P-values for family-based association testing with sequence data

Area of Science:

• Genetics
• Statistical Genetics
• Bioinformatics

Background:

• Family-based studies are crucial for identifying rare causal variants due to enrichment in affected families and robustness to population substructure.
• Existing methods like burden and non-burden tests struggle with rare variants due to dilution from neutral variants.
• Detecting rare causal variants is challenging using single-locus tests.

Purpose of the Study:

• To develop a novel statistical method for identifying rare causal variants in family-based studies.
• To improve the robustness of genetic association tests against neutral variants.
• To enhance the power of detecting rare causal variants, particularly for dichotomous traits.

Main Methods:

• Proposed an 'adaptive combination of P-values method' (ADA) that combines per-site P-values of likely causal variants.
• Discarded variants with large P-values, presumed to be neutral, from the combined statistic.
• Validated the method through extensive simulation studies and application to Genetic Analysis Workshop 17 data.

Main Results:

• The ADA method demonstrated increased robustness to the inclusion of neutral variants compared to existing methods.
• This robustness is particularly advantageous when analyzing dichotomous traits in family-based studies.
• Performance was evaluated using simulated data and real sequence data from the 1000 Genomes Project.

Conclusions:

• The adaptive combination of P-values method (ADA) offers improved power for detecting rare causal variants in family studies.
• ADA is especially beneficial for dichotomous trait analyses due to its robustness against neutral variants.
• Limitations include computational intensity, requirement for pedigree data, and inapplicability to unrelated controls.