Clustering by genetic ancestry using genome-wide SNP data.

Area of Science:

• Genetics
• Bioinformatics
• Population Genetics

Background:

• Population stratification in genome-wide association studies (GWAS) can lead to spurious genetic associations due to differing ancestral backgrounds between cases and controls.
• Principal components analysis (PCA) is commonly used to detect and adjust for population substructure, but genetic matching offers an alternative.
• Effective genetic matching requires well-defined population strata for accurate case and control selection.

Purpose of the Study:

• To develop a novel algorithm for clustering individuals based on ancestral backgrounds derived from PCA.
• To evaluate the algorithm's effectiveness in reducing population stratification bias using real and simulated data.
• To compare the power of this matching approach against traditional PCA adjustment methods.

Main Methods:

• Developed a new algorithm to cluster individuals into ancestral groups using principal components from genome-wide data.
• Applied the algorithm to both simulated and real genetic datasets.
• Utilized the assigned clusters for genetic matching of cases and controls.

Main Results:

• The novel clustering algorithm effectively reduces population stratification bias when matching cases and controls.
• Simulations indicate that this matching method can achieve higher statistical power than adjustment using principal components in specific scenarios.
• The algorithm demonstrates robust performance on both simulated and real-world genetic data.

Conclusions:

• Matching cases and controls using the algorithm's cluster assignments significantly mitigates population stratification bias.
• This approach yields cleaner datasets, facilitating the development of genetic prediction models without the need for ancestry adjustment variables.
• Cluster assignments enable the estimation of genetic heterogeneity by analyzing cluster-specific effects.