Penalized canonical correlation analysis to quantify the association between gene expression and DNA markers

Area of Science:

• Genomics
• Statistical Genetics
• Bioinformatics

Background:

• Gene expression levels exhibit inter-individual variation influenced by DNA markers.
• Associating multiple gene expression and DNA marker variables requires multivariate techniques to account for co-regulation.
• Canonical correlation analysis (CCA) is a suitable multivariate technique for such analyses.

Purpose of the Study:

• To adapt the elastic net, a penalized regression method, for use in canonical correlation analysis.
• To enhance the interpretability of canonical components by reducing the number of variables.
• To effectively group co-regulating genes within the same canonical components.

Main Methods:

• Adaptation of the elastic net, a penalized variable selection approach, to canonical correlation analysis.
• Application of the adapted method to situations where the number of variables significantly exceeds the number of subjects.

Main Results:

• The adapted elastic net significantly reduces the number of variables within canonical components with minimal information loss.
• Co-regulating genes are effectively grouped into the same canonical components, improving biological interpretability.
• The method demonstrates robust performance even when the number of variables greatly exceeds the number of subjects.

Conclusions:

• The adapted elastic net provides an effective and interpretable approach for multivariate analysis of gene expression and DNA marker data.
• This method facilitates the identification of co-regulating gene networks and is particularly useful for high-dimensional genomic datasets.