Discovering the gene-brain-behavior link in autism via generative machine learning.

Area of Science:

• Neuroscience
• Genetics
• Medical Imaging

Background:

• Autism diagnosis is primarily behavioral, despite a known genetic component.
• Understanding the gene-brain-behavior link in autism is complex due to variability.
• A genetics-first approach could revolutionize autism research and treatment.

Purpose of the Study:

• To introduce a novel 3D transport-based morphometry (TBM) technique for autism research.
• To identify structural brain changes associated with genetic copy number variation (CNV) at the 16p11.2 region.
• To establish a genetics-first approach for diagnosing neurodevelopmental disorders.

Main Methods:

• Utilized 3D transport-based morphometry (TBM) to analyze brain structure.
• Applied TBM to data from the Simons Variation in Individuals Project.
• Identified and visualized specific brain endophenotypes linked to 16p11.2 CNV.

Main Results:

• Achieved 89-95% accuracy in predicting 16p11.2 CNV from brain images alone using identified endophenotypes.
• Demonstrated dose-dependent brain changes in deletion and duplication carriers.
• Found endophenotypes correlate with articulation disorders and intelligence quotient variability.

Conclusions:

• TBM is a powerful tool for discovering brain endophenotypes in neurodevelopmental disorders.
• This genetics-first approach can accelerate precision medicine for autism and related conditions.
• The findings contribute to a deeper understanding of neurodiversity and genetic influences on brain structure.