Multiple Linear Regression Allows Weighted Burden Analysis of Rare Coding Variants in an Ethnically Heterogeneous...

Area of Science:

• Genetics
• Statistical Genetics
• Bioinformatics

Background:

• Weighted burden analysis is a method for identifying genes associated with phenotypes using exome-sequenced data.
• Existing methods face challenges when applied to quantitative phenotypes and ethnically diverse datasets.

Purpose of the Study:

• To adapt weighted burden analysis for quantitative phenotypes in large, ethnically heterogeneous exome-sequenced datasets.
• To evaluate the method's performance and its ability to identify plausible gene associations.

Main Methods:

• Developed a weighted burden score incorporating variant frequency and predicted effect.
• Applied the score within a linear regression framework, including population principal components as covariates.
• Utilized a dataset of 49,790 UK Biobank exome-sequenced subjects with body mass index as the phenotype.

Main Results:

• The analysis, after correcting for population structure with 20 principal components, identified plausible gene associations with body mass index, including LYPLAL1 and NSDHL.
• The inclusion of principal components effectively corrected an inflated test statistic observed without them.
• The method demonstrated robustness in an ethnically diverse dataset.

Conclusions:

• The adapted weighted burden analysis is effective for gene-based rare variant association studies in large, heterogeneous cohorts.
• This approach allows for the inclusion of all ethnic groups, preventing data loss.
• The method facilitates the discovery of genes associated with quantitative traits in diverse populations.