ADAPTIVE TESTING OF SNP-BRAIN FUNCTIONAL CONNECTIVITY ASSOCIATION VIA A MODULAR NETWORK ANALYSIS

Area of Science:

• Neuroscience
• Genetics
• Biostatistics

Background:

• Large-scale brain functional networks are high-dimensional and noisy, hindering analysis and interpretation.
• Brain modularity alterations are linked to various neurological disorders.
• Existing methods for brain network estimation struggle with module extraction.

Purpose of the Study:

• To adapt the Weighted Gene Co-expression Network Analysis (WGCNA) framework for identifying modular structures in resting-state fMRI (rs-fMRI) data.
• To develop and apply a novel adaptive statistical test for high-dimensional settings to assess associations between genetic variants and brain network modules.
• To investigate the relationship between genetic variants (APOE4, SNPs) and brain functional network modules using ADNI data.

Main Methods:

• Adapted WGCNA framework for rs-fMRI data to identify modular structures.
• Utilized topological overlap matrix (TOM) elements within hierarchical clustering for module identification.
• Developed and applied a new adaptive proportional odds model (POM) test suitable for high-dimensional data (p > n and p < n).
• Tested associations between genetic variants and whole-brain/subcomponent functional networks using connectivity measures on ADNI data.

Main Results:

• Identified several distinct brain modules within the control cohort.
• Observed marginal associations between some identified modules and the APOE4 genetic variant.
• Found marginal associations between certain modules and other single nucleotide polymorphisms (SNPs).

Conclusions:

• The adapted WGCNA framework effectively identifies modular structures in brain functional networks from rs-fMRI data.
• The novel adaptive POM test is applicable to high-dimensional genetic association studies of brain networks.
• Preliminary findings suggest potential links between genetic variants and brain network modularity, warranting further investigation in larger cohorts.