Clustering and principal-components approach based on heritability for mapping multiple gene expressions

Area of Science:

• Genetics
• Bioinformatics
• Statistical Genetics

Background:

• Single-trait genetic analysis is computationally intensive for thousands of phenotypes, requiring extensive multiple comparison adjustments.
• Traditional multivariate genetic linkage analysis is limited to a few phenotypes, making it unsuitable for high-dimensional data.
• Standard clustering and principal-component analysis (PCA) methods cannot effectively utilize family structure information, leading to data loss in genetic studies.

Purpose of the Study:

• To develop and apply a novel clustering method that incorporates family structure for high-dimensional phenotype data.
• To reduce data dimensionality and identify shared genetic contributions across multiple traits, specifically gene expression levels.
• To improve the power of genetic linkage analysis by integrating family information.

Main Methods:

• A new clustering method was developed to exploit family structure in genetic data.
• Principal-component analysis (PCA) approaches, including a penalized version, were used to combine phenotypes based on heritability.
• Genome-wide multipoint linkage analysis was performed on individual and combined traits.

Main Results:

• The proposed clustering method successfully utilized family structure information for 29 gene expression levels.
• Linkage analysis identified two previously reported peaks on chromosomes 14 and 20.
• Methods incorporating family structure information yielded stronger linkage evidence compared to standard approaches.

Conclusions:

• Novel clustering and PCA methods that account for family structure improve genetic linkage analysis for high-dimensional phenotypes.
• These methods are crucial for accurately mapping genetic contributions in studies with complex family data.
• The findings highlight the importance of utilizing family structure to maximize information in genetic studies.