A multivariate to multivariate approach for voxel-wise genome-wide association analysis

Area of Science:

• Neuroimaging Genetics
• Computational Neuroscience
• Statistical Genetics

Background:

• Investigating genetic influences on brain structure and function requires analyzing vast imaging-genetics datasets.
• Voxel-wise genome-wide association analysis (GWAS) involves examining trillions of single nucleotide polymorphism (SNP)-voxel pairs.
• Understanding polygenic and pleiotropic networks in brain imaging traits is crucial.

Purpose of the Study:

• To develop a systematic method for identifying organized association patterns between SNPs and brain voxels.
• To detect latent SNP-voxel bi-cliques and establish a statistical inference model.
• To uncover genetic loci influencing white matter integrity.

Main Methods:

• Proposed a bi-clique graph structure to represent SNP-voxel associations.
• Developed computational strategies for detecting latent SNP-voxel bi-cliques.
• Implemented a statistical inference model with theoretical accuracy guarantees.
• Applied the method to whole-genome genetic and white matter integrity data from the Human Connectome Project (1052 participants).

Main Results:

• Successfully identified organized association patterns between genetic variations and brain imaging traits.
• Demonstrated the method's accuracy through extensive simulation studies.
• Discovered multiple genetic loci significantly associated with white matter integrity in the splenium and genu of the corpus callosum.

Conclusions:

• The proposed bi-clique graph approach offers a systematic way to analyze complex imaging-genetics data.
• This method effectively identifies genetic influences on specific brain structures like the corpus callosum.
• The findings contribute to understanding the genetic architecture of brain white matter integrity.