An eigenvalue ratio approach to inferring population structure from whole genome sequencing data

Area of Science:

• Population genetics
• Medical genetics
• Genomics

Background:

• Whole genome sequencing data offers rich population structure insights.
• Traditional methods for principal component analysis (PCA) may fail with large genomic datasets.
• Existing PCA methods struggle with high variant numbers and linkage disequilibrium in sequencing data.

Purpose of the Study:

• To propose a novel method, ERStruct, for selecting informative principal components (PCs) from population sequencing data.
• To address the limitations of traditional PCA methods when applied to whole genome sequencing data.
• To improve the accuracy of population structure inference using genetic sequencing data.

Main Methods:

• Developed ERStruct, a new method for determining the number of top PCs.
• Utilized the ratio of consecutive eigenvalues as a robust test statistic.
• Approximated the null distribution of the test statistic using random matrix theory.

Main Results:

• ERStruct effectively determines the number of informative PCs from sequencing data.
• The method overcomes limitations of traditional approaches, such as the Tracy-Widom test.
• Demonstrated robust performance on simulated data and real-world datasets (HapMap 3, 1000 Genomes).

Conclusions:

• ERStruct provides a reliable approach for principal component selection in population genetics using sequencing data.
• The method enhances the inference of population structure from large-scale genetic datasets.
• ERStruct offers a significant advancement for population and medical genetics research.