Multi-view gene panel characterization for spatially resolved omics

Area of Science:

• Genomics
• Bioinformatics
• Molecular Biology

Background:

• Spatially resolved transcriptomics offers cellular resolution but relies on pre-selected gene panels.
• Current gene panel design often prioritizes cell type identification over other crucial factors.
• Effective panel design requires consideration of transcriptional variation, pathway coverage, and gene redundancy.

Purpose of the Study:

• To develop a framework for comprehensive gene panel characterization and optimization for spatial transcriptomics.
• To introduce panelScope for holistic panel comparison and panelScope-OA for automated panel optimization.
• To provide quantitative, multi-dimensional insights for designing tailored gene panels.

Main Methods:

• Developed panelScope, a platform for characterizing gene panels from multiple perspectives.
• Created panelScope-OA, a genetic algorithm integrating characterization metrics for automated panel optimization.
• Applied panelScope and panelScope-OA to analyze nine gene panels across four datasets.

Main Results:

• Computationally designed gene panels demonstrated competitive performance in capturing major cell types.
• Manual curation showed advantages in identifying and capturing minor cell types.
• panelScope and panelScope-OA provided quantitative insights for panel design.

Conclusions:

• The developed framework offers a multi-dimensional approach to gene panel design for spatial transcriptomics.
• Automated and characterization tools can support the creation of customized gene panels.
• Balancing computational design with expert curation may yield optimal gene panels for diverse research needs.