A robust statistical method for Genome-wide association analysis of human copy number variation

TL;DR

This paper introduces a robust statistical method for GWAS of human CNV data that effectively handles batch effects and heterogeneity, improving detection power and reducing false discoveries.

Contribution

A novel robust statistical approach for CNV-based GWAS that incorporates an empirical Bayes rule and provides theoretical guarantees, addressing limitations of traditional methods.

Findings

Method outperforms traditional approaches in simulations

Effective in real data analysis with reduced false positives

Robust to batch effects and heterogeneity

Abstract

Conducting genome-wide association studies (GWAS) in copy number variation (CNV) level is a field where few people involves and little statistical progresses have been achieved, traditional methods suffer from many problems such as batch effects, heterogeneity across genome, leading to low power or high false discovery rate. We develop a new robust method to find disease-risking regions related to CNV's disproportionately distributed between case and control samples, even if there are batch effects between them, our test formula is robust to such effects. We propose a new empirical Bayes rule to deal with overfitting when estimating parameters during testing, this rule can be extended to the field of model selection, it can be more efficient compared with traditional methods when there are too much potential models to be specified. We also give solid theoretical guarantees for our proposed method, and demonstrate the effectiveness by simulation and realdata analysis.