Identifying genetic associations with MRI-derived measures via tree-guided sparse learning.

Area of Science:

• Neuroimaging Genetics
• Computational Biology
• Statistical Genetics

Background:

• Imaging genetics studies commonly use regression analysis to associate single nucleotide polymorphisms (SNPs) with quantitative traits (QTs).
• Feature selection methods like Lasso are employed for high-throughput genetic data, but often ignore the hierarchical structure of SNPs.
• Existing methods may miss important biological information due to the underutilization of genomic structural relationships.

Purpose of the Study:

• To develop and evaluate a novel tree-guided sparse learning (TGSL) method for identifying associations between SNPs and MRI-derived measures.
• To incorporate prior hierarchical information of SNPs (gene-based and linkage disequilibrium-based) into a sparse learning framework.
• To enhance the accuracy of SNP association detection and improve biological interpretation in imaging genetics.

Main Methods:

• Proposed a tree-guided sparse learning (TGSL) model that integrates SNP hierarchical structures.
• Utilized two types of hierarchical information: grouping by gene and grouping by linkage disequilibrium (LD) clusters.
• Applied the TGSL method to analyze associations between SNPs and MRI measures in the Alzheimer's Disease Neuroimaging Initiative (ADNI) database.

Main Results:

• The TGSL method demonstrated superior prediction performance for MRI measures, specifically the left and right hippocampal formation.
• The approach successfully identified informative SNPs associated with MRI measures.
• TGSL outperformed reference methods in both predictive accuracy and the identification of biologically relevant genetic markers.

Conclusions:

• The proposed TGSL method effectively utilizes SNP hierarchical structures for improved feature selection in imaging genetics.
• This approach enhances the identification of genetic associations with neuroimaging phenotypes.
• The findings suggest TGSL is a powerful tool for advancing our understanding of the genetic underpinnings of brain structure and neurological disorders.