[Paper Review] The human brain's network architecture is genetically encoded by modular pleiotropy

Maxwell A. Bertolero, Ann S. Blevins+6 more|arXiv (Cornell University)|May 18, 2019

Functional Brain Connectivity Studies56 references17 citations

TL;DR

This study reveals that the human brain's network architecture is genetically encoded through modular pleiotropy, where distinct gene sets govern the connectivity of modular brain networks, while connector hubs—integrative regions—rely on pleiotropic genes from non-overlapping sets. The findings demonstrate that brain network communities correspond to genetic boundaries, offering a new framework for understanding brain development, evolution, and disease through genetic mapping.

ABSTRACT

For much of biology, the manner in which genotype maps to phenotype remains a fundamental mystery. The few maps that are known tend to show modular pleiotropy: sets of phenotypes are determined by distinct sets of genes. One key map that has evaded discovery is that of the human brain's network architecture. Here, we determine the form of this map for gene coexpression and single nucleotide polymorphisms. We discover that mostly non-overlapping sets of genes encode the connectivity of brain network modules (or so-called communities), suggesting that brain network communities demarcate genetic transitions. We find that these clean boundaries break down at connector hubs, whose integrative connectivity is encoded by pleiotropic genes from mostly non-overlapping sets. Broadly, this study opens fundamentally new directions in the study of genetic encoding of brain development, evolution, and disease.

Motivation & Objective

• To uncover the genetic basis underlying the human brain's network architecture.
• To determine whether gene expression and genetic variation map to specific brain network modules.
• To investigate whether genetic encoding of brain networks follows modular pleiotropy, with non-overlapping gene sets for distinct modules.
• To examine the role of connector hubs in genetic encoding and their deviation from modular boundaries.
• To establish a foundational map linking genotype to brain network phenotypes, advancing understanding of brain development and disease.

Proposed method

• Applied graph theory to structural and functional brain networks derived from neuroimaging data to identify modular communities and connector hubs.
• Used gene coexpression networks from the Allen Human Brain Atlas to map gene expression patterns across brain regions.
• Mapped single nucleotide polymorphisms (SNPs) from the Philadelphia Neurodevelopmental Cohort to brain network topology.
• Employed statistical methods to test for non-overlapping gene sets encoding distinct network modules, assessing genetic modularity.
• Quantified pleiotropy by identifying genes associated with multiple network features, particularly in connector hubs.
• Used permutation testing and network-based statistics to validate the significance of genetic associations with network structure.

Experimental results

Research questions

• RQ1 How are distinct brain network modules genetically encoded in terms of gene expression and genetic variation?
• RQ2 To what extent do gene sets associated with different network modules show non-overlapping genetic contributions?
• RQ3 How do connector hubs—highly connected regions—differ genetically from other network modules in terms of gene pleiotropy?
• RQ4 Is there evidence of modular pleiotropy in the genetic architecture of human brain networks?
• RQ5 Can we identify a systematic mapping from genotype to brain network topology using coexpression and SNP data?

Key findings

• Distinct, mostly non-overlapping gene sets encode the connectivity of different brain network modules, indicating modular genetic control.
• Connector hubs, which integrate multiple modules, are genetically encoded by pleiotropic genes drawn from non-overlapping sets, suggesting a unique genetic mechanism for integration.
• The genetic architecture of brain networks reveals clean boundaries between modules, with minimal genetic overlap across modules.
• Genetic variation (SNPs) and gene coexpression patterns both show strong associations with modular network structure, supporting a robust genetic encoding of topology.
• The breakdown of genetic modularity at connector hubs indicates a transition zone where pleiotropy enables integrative function.
• The study establishes a direct link between genetic variation and brain network architecture, providing a foundational map for future studies in neurogenomics.