Adjusting for population stratification in a fine scale with principal components and sequencing data

Area of Science:

• Population genetics
• Genomic data analysis
• Human evolution

Background:

• Population stratification is crucial for understanding human evolution and avoiding errors in genetic association studies.
• Next-generation sequencing offers new possibilities for fine-scale population structure analysis, but presents challenges.
• Minor allele frequencies (MAFs) of variants can have varying confounding effects on population stratification.

Purpose of the Study:

• To compare the effectiveness of Principal Component Analysis (PCA) and Spectral Dimensional Reduction (SDR) in detecting and adjusting for population stratification.
• To investigate the performance of different variant types and sets in controlling stratification effects across various genetic testing scenarios.
• To evaluate methods for fine-scale population structure analysis using low-coverage sequencing data.

Main Methods:

• Utilized low-coverage sequencing data from the 1000 Genomes Project.
• Compared Principal Component Analysis (PCA) with Spectral Dimensional Reduction (SDR), a spectral clustering technique.
• Assessed the performance of different variant sets (all, common, rare) for stratification adjustment and testing.

Main Results:

• Principal components derived from all or common variants were most effective in controlling population stratification.
• Rare variants did not significantly improve the control of stratification, contrary to some expectations.
• Spectral Dimensional Reduction (SDR) demonstrated greater robustness than PCA, particularly when analyzing rare variants.

Conclusions:

• Common variants are generally most effective for Principal Component Analysis (PCA)-based population stratification control.
• Spectral Dimensional Reduction (SDR) offers a more robust alternative to PCA, especially for rare variant analysis.
• The choice of variants significantly impacts the accuracy of population stratification detection and adjustment in genetic studies.