Using canonical correlation analysis to discover genetic regulatory variants

Area of Science:

• Genetics
• Bioinformatics
• Systems Biology

Background:

• Identifying genetic associations with gene expression is crucial for understanding gene regulation and disease mechanisms.
• Traditional methods using expression data as phenotypes result in numerous multiple comparisons, potentially reducing statistical power.
• Investigating genetic variants that regulate gene expression is a key area in genomics.

Purpose of the Study:

• To evaluate the efficacy of canonical correlation analysis (CCA) for identifying genetic variants associated with gene expression.
• To assess CCA's potential in reducing multiple comparisons and increasing power in genomewide association studies.
• To explore CCA as a method for discovering regulatory variants.

Main Methods:

• Applied canonical correlation analysis to partitioned genomewide data.
• Utilized simulations to compare CCA with standard pairwise univariate regression.
• Demonstrated the approach using data from the Childhood Asthma Management Program (CAMP).

Main Results:

• Canonical correlation analysis showed higher power than univariate regression for detecting single nucleotide polymorphisms (SNPs) linked to gene expression, particularly with low heritability.
• The power advantage of CCA was more pronounced under a recessive genetic model.
• The method was successfully applied to real-world genetic and gene expression data.

Conclusions:

• CCA offers a powerful approach to discover genetic associations with gene expression.
• This method effectively reduces the burden of multiple comparisons inherent in traditional analyses.
• The findings provide insights into the complex genotype-gene expression relationships and regulatory variants.