BrainGENIE: The Brain Gene Expression and Network Imputation Engine.

Area of Science:

• Neuroscience
• Computational Biology
• Genomics

Background:

• In vivo human brain tissue analysis is challenging and ethically complex.
• Existing methods for predicting brain gene expression have limitations.

Purpose of the Study:

• To develop a computational method, Brain Gene Expression and Network-Imputation Engine (BrainGENIE), to predict brain tissue-specific gene expression from peripheral blood transcriptomes.
• To compare BrainGENIE's performance against existing transcriptome imputation methods like PrediXcan.

Main Methods:

• Utilized paired blood-brain transcriptomic data from the Genotype-Tissue Expression (GTEx) Project to train BrainGENIE models.
• Trained models to predict gene expression in ten distinct brain regions using whole-blood gene expression profiles.
• Compared BrainGENIE's predictive accuracy with PrediXcan.

Main Results:

• BrainGENIE significantly predicted brain tissue-specific gene expression for 2947-11,816 genes (FDR-adjusted p < 0.05).
• Identified numerous transcripts unpredictably by other methods.
• BrainGENIE better recapitulated diagnosis-related gene expression changes in autism, bipolar disorder, and schizophrenia compared to direct blood correlations and PrediXcan predictions.

Conclusions:

• BrainGENIE provides a powerful, non-invasive tool for studying brain gene expression.
• The software toolset facilitates BrainGENIE implementation, opening new research avenues in neuroscience and psychiatric disorders.
• BrainGENIE complements and surpasses existing transcriptome imputation tools, offering biologically meaningful predictions.