A unified framework for identification of cell-type-specific spatially variable genes in spatial transcriptomic...

Area of Science:

• Genomics
• Bioinformatics
• Systems Biology

Background:

• Spatial transcriptomics (ST) studies require methods to identify cell-type-specific spatially variable genes (SVGs) within tissue context.
• Existing methods may not adequately account for cell type composition or platform effects in ST data.

Purpose of the Study:

• To present a unified framework, the Mixture of Mixed Models (MMM), for identifying cell-type-specific SVGs.
• To directly model RNA count data while accounting for cell type composition and correcting for platform effects.

Main Methods:

• Developed the Mixture of Mixed Models (MMM) framework.
• Applied MMM to eight publicly available ST datasets and a high-resolution Xenium human breast cancer dataset.
• Integrated MMM results with genome-wide association studies (GWAS).

Main Results:

• MMM effectively identifies cell-type-specific SVGs across diverse ST datasets and technologies.
• Identified SVGs in mouse brain ST data show significant heritability enrichment for brain-related phenotypes.
• Analysis of breast cancer ST data suggests SVGs contribute to cell-cell communication and regulate the tissue microenvironment.
• Demonstrated MMM's versatility in analyzing 3D ST data.

Conclusions:

• MMM is a robust and effective tool for identifying cell-type-specific SVGs in ST studies.
• Cell-type-specific SVGs are crucial for understanding complex traits, diseases, and tissue microenvironment regulation.
• MMM facilitates deeper insights into biological systems using ST data, including 3D analyses.