Multivariate regression analysis of distance matrices for testing associations between gene expression patterns and...

Area of Science:

• Genomics
• Bioinformatics
• Statistical Modeling

Background:

• High-dimensional genomic data analysis, such as gene expression, relies on calculating sample similarity or distance.
• An N x N similarity matrix is typically formed to represent pairwise sample correlations based on gene expression levels.
• Conventional methods often involve data reduction and cluster analysis of these similarity matrices.

Purpose of the Study:

• To present an alternative multivariate analysis method for genomic similarity matrices.
• To relate predictor variables of samples to variations within the pairwise similarity/distance values.
• To avoid the need for dimension reduction in similarity matrix analysis.

Main Methods:

• Utilizes traditional linear models to analyze similarity matrices.
• Relates predictor variables to pairwise similarity/distance values.
• Applies a multivariate approach to genomic data analysis.

Main Results:

• The proposed method avoids dimension reduction of similarity matrices.
• It facilitates the assessment of relationships between genes and additional sample information.
• The technique is versatile for analyzing individual or grouped genes.

Conclusions:

• The multivariate method offers a powerful alternative for analyzing genomic similarity matrices.
• It enhances the understanding of relationships between genomic data and sample characteristics.
• The technique is applicable to various high-dimensional data types and gene subset analyses.