Principal component of explained variance: An efficient and optimal data dimension reduction framework for...

Area of Science:

• Genomics
• Statistical Genetics
• Bioinformatics

Background:

• The genomics era generates high-dimensional data, necessitating dimension-reduction techniques for biological association studies.
• Principal Component of Heritability (PCEV) is a dimension-reduction method that maximizes variance explained by covariates but has faced computational challenges.
• Existing methods struggle with the computational complexity of analyzing high-dimensional biological data.

Purpose of the Study:

• To revisit and rename the Principal Component of Heritability as the Principal Component of Explained Variance (PCEV).
• To propose a general analytical PCEV framework that is computationally simple and retains the original method's strengths.
• To extend PCEV applications to high-dimensional outcomes and provide efficient testing procedures.

Main Methods:

• Developed a general analytical framework for the Principal Component of Explained Variance (PCEV).
• Implemented computationally simple strategies for high-dimensional outcome data.
• Introduced exact and asymptotic testing procedures to reduce computational cost.

Main Results:

• The proposed PCEV framework is computationally efficient for high-dimensional data.
• The method optimizes statistical power for detecting associations with covariates.
• Simulations demonstrate the effectiveness of the PCEV approach.

Conclusions:

• The enhanced PCEV framework offers a computationally feasible and powerful tool for analyzing high-dimensional genomics data.
• PCEV is applicable to various fields, including epigenetics and brain imaging.
• This approach simplifies complex biological data analysis for covariate association testing.