Multiscale comparative connectomics.

Area of Science:

• Neuroscience
• Computational Biology
• Graph Theory

Background:

• The brain connectome, mapping neural connections, offers insights into neurobiological phenotypes and neurological disorders.
• Current computational methods for analyzing multi-subject connectome data are often limited, failing to capture complex network topologies or analyze multiple datasets simultaneously.

Purpose of the Study:

• To introduce novel, robust, and interpretable statistical hypothesis tests for analyzing multiple connectomes.
• To enable the discovery of hierarchical brain structures that correlate with phenotypic profiles across different scales of network topology.

Main Methods:

• Development of statistical hypothesis tests based on random graph models for multi-subject connectomics.
• Simultaneous analysis of multiple connectomes across various network topology scales.
• Validation through extensive simulation studies and real-data experiments on mouse models.

Main Results:

• The proposed methods demonstrate superiority over current state-of-the-art connectomics techniques.
• Successfully uncovered latent information in multi-subject connectomics data from genetically distinct mouse strains.
• Identified connective correlates of neurological phenotypes not captured by other methods.

Conclusions:

• The novel statistical tests provide a more rigorous and comprehensive approach to connectomics analysis.
• These methods facilitate the discovery of hierarchical brain structures and their relationship to phenotypes.
• Offers valuable insights into the neurobiological underpinnings of neurological conditions.