Association testing for next-generation sequencing data using score statistics

Area of Science:

• Genomics
• Statistical Genetics
• Bioinformatics

Background:

• Large-scale sequencing studies are crucial for identifying genetic variants linked to diseases.
• Genotype calling uncertainty in sequencing data can reduce statistical power and introduce false signals.
• Existing methods to address genotype uncertainty are often computationally intensive for large datasets.

Purpose of the Study:

• To develop a computationally feasible statistical method for genetic association testing in next-generation sequencing data.
• To improve the power of association studies by explicitly accounting for genotype classification uncertainty.
• To provide a flexible framework applicable to both quantitative and discrete phenotypes, including case-control studies.

Main Methods:

• Utilizes a score statistic for the joint likelihood of observed phenotypes and sequencing data.
• Incorporates genotype uncertainty using posterior probabilities of genotypes derived from sequencing data.
• Employs a generalized linear model framework to accommodate quantitative and discrete phenotypes, allowing for covariates.

Main Results:

• The proposed score statistic method demonstrates higher statistical power compared to methods relying solely on called genotypes.
• The approach effectively accounts for genotype classification uncertainty, mitigating spurious association signals.
• The method is computationally feasible for large-scale sequencing studies.

Conclusions:

• The score statistic approach offers a powerful and computationally efficient solution for association testing with next-generation sequencing data.
• This method enhances the reliability of genetic association studies by properly handling genotype uncertainty.
• The generalized linear model framework ensures broad applicability across various study designs and phenotype types.