Batch effects in the BRLMM genotype calling algorithm influence GWAS results for the Affymetrix 500K array.

Area of Science:

• Genetics
• Bioinformatics
• Statistical Genetics

Background:

• Genome-wide association studies (GWASs) are vital for identifying genetic variants associated with diseases.
• Accurate genotype calling is essential to minimize Type I and Type II errors in GWASs.
• Genotype calling errors can introduce bias, impacting the reliability of association study findings.

Purpose of the Study:

• To investigate the impact of batch effects on genotype calling accuracy using the Affymetrix 500K array.
• To assess how batch size and composition influence downstream association analysis in GWASs.
• To understand the contribution of these factors to the lack of reproducibility in GWASs.

Main Methods:

• Utilized data from the Wellcome Trust Case Control Consortium and UK Blood Services (NBS).
• Genotyped 1991 coronary artery disease (CAD) cases and 1500 controls on the Affymetrix 500K array.
• Applied the Bayesian Robust Linear Model with Mahalanobis distance classifier (BRLMM) algorithm, varying batch sizes and compositions.

Main Results:

• Batch composition and size introduced 2-3% discordance in quality control and statistical analysis.
• Changes in batch size were the primary driver of differential single-nucleotide polymorphism (SNP) results.
• An interactive effect between batch size and composition was observed, contributing to discordant results for significantly associated loci.

Conclusions:

• Batch effects related to size and composition in genotype calling algorithms can significantly impact GWAS results.
• These variations can lead to a lack of reproducibility across different GWAS.
• Careful consideration of batch parameters is necessary for robust and reproducible genetic association studies.