A Log-Ratio Biplot Approach for Exploring Genetic Relatedness Based on Identity by State

Area of Science:

• Genomics
• Population Genetics
• Bioinformatics

Background:

• Detecting cryptic relatedness in large population-based cohorts is vital for genome research.
• Current methods using allele sharing statistics in 2D scatterplots have limitations in capturing higher dimensionality and compositional data.

Purpose of the Study:

• To develop novel biplot methodology based on log-ratio principal component analysis (LRPCA) for exploring relatedness in genetic databases.
• To overcome the restrictions of traditional methods and provide new graphical tools for relationship exploration.

Main Methods:

• Developed biplot methodology using log-ratio principal component analysis.
• Applied the method to datasets from the 1,000 Genomes Project and Genomes For Life-GCAT Project.
• Conducted a simulation study comparing LRPCA with classical plots using identity by state.

Main Results:

• The LRPCA approach provides new graphical tools for exploring relatedness in homogeneous populations.
• The method can be applied iteratively to identify more remote relationships.
• In simulations, LRPCA outperformed classical graphics in classifying relationships up to the fourth degree in non-inbred homogeneous populations using 35,000 variants.

Conclusions:

• Log-ratio biplot methodology offers a powerful and more discriminatory approach for detecting cryptic relatedness compared to classical methods.
• LRPCA, combined with discriminant analysis, can accurately classify familial relationships in large genetic datasets.