CellGO: a novel deep learning-based framework and webserver for cell-type-specific gene function interpretation.

Area of Science:

• Computational Biology
• Genomics
• Bioinformatics

Background:

• Interpreting omics data requires understanding gene function within specific biological contexts.
• Current pathway analysis tools often lack tissue or cell-type specificity.
• Identifying cell-type-specific gene activity is crucial for biological insights.

Purpose of the Study:

• To develop a novel computational tool, CellGO, for cell-type-specific pathway analysis.
• To address the limitations of existing tools in incorporating biological context.
• To enable accurate identification of active pathways at the single-cell level.

Main Methods:

• Utilized visible neural networks (VNNs) and single-cell gene expression data.
• Developed a scoring system for cell-type-specific gene-pathway activity.
• Integrated pathway community analysis and identification of driver genes and cell types.

Main Results:

• CellGO successfully identified cell-type-specific active pathways from mixed cell-type markers.
• Demonstrated superiority in identifying cell types and pathways affected by gene knockouts.
• Showcased ability to infer cell-type-specific disease pathogenesis for risk genes.

Conclusions:

• CellGO provides a versatile and accessible solution for cell-type-specific pathway analysis.
• The tool enhances biological interpretation of omics data.
• CellGO is available as a Python package with a web interface for broad research application.