Genome-wide Variant-based Study of Genetic Effects with the Largest Neuroanatomic Coverage

Jin Li, Wenjie Liu, Huang Li, Feng Chen, Haoran Luo, Peihua Bao, Yanzhao Li, Hailong Jiang, Yue Gao, Hong Liang, Shiaofen FangResearch 2020 · DOI

Abstract

Brain imaging genetics explores how genetic variations influence brain structure and function, particularly in neurodegenerative diseases like Alzheimer’s. This study introduces an informatics strategy to identify single nucleotide polymorphisms (SNPs) and SNP sets with the most extensive neuroanatomic coverage at both voxel and region-of-interest (ROI) levels. By implementing a Genetic Algorithm (GA), researchers efficiently identified high-order SNP sets that influence broad brain regions, overcoming the computational limitations of exhaustive searches. Results highlight specific variants and combinations that correlate with systemic brain changes, offering a more comprehensive view of the genetic architecture underlying brain anatomy.

Introduction: The Challenge of Imaging Genetics

Imaging genetics aims to bridge the gap between our DNA and the physical structure of the brain. While traditional Genome-Wide Association Studies (GWAS) often focus on isolated regions of interest (ROIs), such as the hippocampus, brain diseases like Alzheimer’s typically manifest as widespread changes across interconnected networks. Identifying the genetic drivers of these systemic changes requires analyzing "neuroanatomic coverage"—the extent to which a genetic variant affects the brain's overall anatomy.

However, searching for combinations of multiple SNPs that maximize this coverage presents a "combinatorial explosion." For instance, finding the optimal set of three SNPs from a pool of thousands involves hundreds of millions of possibilities, making exhaustive computational searches impractical for discovery.

Methods: A Multi-Level Informatics Strategy

The researchers developed a framework to rank genetic effects based on their breadth of influence across the brain using data from the Alzheimer’s Disease Neuroimaging Initiative (ADNI).

  1. Voxelwise GWAS (vGWAS): Baseline associations were established between 1,784 SNPs and over 271,000 brain voxels.
  2. Ranking Metrics: SNPs were categorized using four primary metrics: Voxel-based coverage (VBC), Voxel-based p-value (VBP), ROI-based coverage (RBC), and ROI-based p-value (RBP).
  3. Genetic Algorithm (GA): To identify 3-SNP sets efficiently, the team employed a GA. This heuristic search mimics natural selection—evolving a population of SNP combinations through crossover and mutation to find the highest "coverage score" without exhaustive testing.
Top individual SNPs ranked by neuroanatomic coverage metrics (VBC, VBP, RBC, RBP).

Key Results: Efficiency and Discovery

The study successfully identified an ensemble of top-performing SNPs and SNP sets. Notably, rs6092321 (in the RTF2 gene) and rs6024860 emerged as variants with the highest individual neuroanatomic coverage, accounting for significant phenotypic variance.

The Genetic Algorithm demonstrated remarkable efficiency. While an exhaustive search for 3-SNP sets would have required over 30 days of computation, the GA reached an optimal solution in less than 20 minutes. This rapid optimization is visualized through the relationship between the algorithm's parameters and the resulting coverage score.

Optimization of the Genetic Algorithm score based on population size and evolution generations.

Beyond individual effects, the study revealed that certain SNPs which appear insignificant in isolation show massive neuroanatomic coverage when grouped. These "team player" variants, found in genes like CNTNAP2 and FAM210B, suggest that much of the "missing heritability" in brain structure may reside in complex genetic interactions. The identified SNP sets also showed significant correlations with clinical features of Alzheimer’s, including memory performance scores.

Conclusion

This research provides a scalable informatics strategy for mapping the genetic influences on brain anatomy. By prioritizing neuroanatomic coverage and utilizing evolutionary algorithms, the framework allows for the discovery of complex genetic sets that drive systemic changes in the brain. Future work will focus on scaling these methods to true genome-wide datasets and validating the biological mechanisms behind these identified associations.

Summary updated

Abstract

Abstract Background: Brain image genetics provides enormous opportunities for examining the effects of geneticvariations on the brain. Many studies have shown that the structure, function, and abnormality (e.g., thoserelated to Alzheimer's disease) of the brain are heritable. However, which genetic variations contribute to thesephenotypic changes is not completely clear. Advances in neuroimaging and genetics have led us to obtaindetailed brain anatomy and genome-wide information. These data offer us new opportunities to identify geneticvariations such as single nucleotide polymorphisms (SNPs) that affect brain structure. In this paper, we performa genome-wide variant-based study, and aim to identify top SNPs or SNP sets which have genetic effects withthe largest neuroanotomic coverage at both voxel and region-of-interest (ROI) levels. Based on the voxelwisegenome-wide association study (GWAS) results, we used the exhaustive search to nd the top SNPs or SNPsets that have the largest voxel-based or ROI-based neuroanatomic coverage. For SNP sets with >2 SNPs, weproposed an efficient genetic algorithm to identify top SNP sets that can cover all ROIs or a specific ROI.Results: We identified an ensemble of top SNPs, SNP-pairs and SNP-sets, whose effects have the largestneuroanatomic coverage. Experimental results on real imaging genetics data show that the proposed geneticalgorithm is superior to the exhaustive search in terms of computational time for identifying top SNP-sets.Conclusions: We proposed and applied an informatics strategy to identify top SNPs, SNP-pairs and SNP-setsthat have genetic effects with the largest neuroanatomic coverage. The proposed genetic algorithm others anefficient solution to accomplish the task, especially for identifying top SNP-sets.
BibTeX
@inproceedings{li2020genome,
  author = {Jin Li and Wenjie Liu and Huang Li and Feng Chen and Haoran Luo and Peihua Bao and Yanzhao Li and Hailong Jiang and Yue Gao and Hong Liang and Shiaofen Fang},
  title = {{Genome-wide Variant-based Study of Genetic Effects with the Largest Neuroanatomic Coverage}},
  booktitle = {Research},
  year = {2020},
  doi = {10.21203/rs.3.rs-33088/v2},
}
Figure