A method to address differential bias in genotyping in large-scale association studies

Area of Science:

• Genetics
• Bioinformatics
• Statistical Genetics

Background:

• High-throughput genotyping in large-scale disease association studies often involves DNA samples from different sources.
• This variability can introduce scoring inaccuracies, leading to differential misclassification and increased false-positive rates.

Purpose of the Study:

• To describe methodological improvements for minimizing biases caused by different DNA sourcing in genetic association studies.
• To enhance the accuracy of genotype calling and reduce false-positive rates.

Main Methods:

• Modification of basic clustering methods for genotype identification from fluorescence intensities.
• Implementation of "fuzzy" genotype calls in association tests to account for call uncertainty.
• Development of a modified calling algorithm that links case and control clustering despite different DNA sourcing.

Main Results:

• A modified calling algorithm effectively links case and control clustering while accommodating different DNA sources.
• Biases from missing data in the presence of varied DNA sourcing can elevate the false-positive rate.
• "Fuzzy" calls effectively address uncertain genotypes that might otherwise be classified as missing.

Conclusions:

• Methodological improvements, particularly a modified clustering algorithm and the use of "fuzzy" calls, can significantly minimize biases in genetic association studies with diverse DNA sources.
• These approaches enhance genotype accuracy and reduce false-positive rates, crucial for reliable disease association findings.