TransBrain: a computational framework for translating brain-wide phenotypes between humans and mice.

Area of Science:

• Neuroscience
• Computational Biology
• Genomics

Background:

• Quantitative cross-species brain research is limited by challenges in bridging preclinical rodent and human studies.
• Existing methods lack accurate homology mapping between mouse and human brain structures.

Purpose of the Study:

• To present TransBrain, a computational framework for bidirectional translation of brain-wide phenotypes between humans and mice.
• To enhance human-mouse homology mapping accuracy and enable quantitative cross-species comparisons.

Main Methods:

• Developed a deep neural network using multimodal human transcriptomics for region-specific cortical and subcortical mapping.
• Constructed a unified cross-species representational space using a graph-based approach incorporating anatomical and connectivity data.
• Applied TransBrain to assess brain organization, infer cognitive functions, and translate molecular insights for autism research.

Main Results:

• Achieved an 89.5% improvement in cortical correspondence using the deep neural network.
• Identified two evolutionarily conserved gradients in the human and mouse cortex.
• Demonstrated utility in quantitative assessment of brain features, cognitive function inference, and autism mechanism translation.

Conclusions:

• TransBrain facilitates accurate, quantitative cross-species comparisons of brain function and organization.
• The framework enables mechanistic investigation of normal and pathological brain functions across species.
• TransBrain advances translational neuroscience by bridging the gap between preclinical models and human studies.