MINGL Quantifies Borders, Gradients, and Heterogeneity in Multicellular Tissue Organization.

Area of Science:

• Spatial Biology
• Computational Biology
• Systems Biology

Background:

• Tissues exhibit complex organizational units with critical interfaces influencing function.
• Current spatial-omics methods often oversimplify cellular neighborhoods, ignoring gradients and heterogeneity.
• Understanding tissue architecture is vital for comprehending health and disease states.

Purpose of the Study:

• To develop a probabilistic framework, MINGL, for quantifying continuous tissue architecture from discrete neighborhood annotations.
• To enable the identification of cellular interfaces, interaction networks, and compositional gradients within tissues.
• To provide a scalable and comparable method for analyzing spatial-omics data across diverse biological contexts.

Main Methods:

• Developed MINGL (Mixture-based Identification of Neighborhood Gradients with Likelihood estimates), a probabilistic framework.
• Modeled cells with multi-membership probabilities across hierarchical organizational units.
• Applied MINGL to spatial-omics datasets from melanoma, healthy intestine, and Barrett's Esophagus.

Main Results:

• MINGL identified innate immune-enriched interfaces at tumor and anatomical boundaries.
• Detected plasma cell niches connecting distinct cellular neighborhoods.
• Quantified sharp and gradual compositional gradient transitions and disease-associated neighborhood remodeling.
• Demonstrated MINGL's ability to unify discrete and continuous representations of tissue organization.

Conclusions:

• MINGL transforms neighborhood assignments into continuous measures of tissue architecture.
• The framework enhances the analysis of cellular interactions and compositional heterogeneity.
• MINGL offers a scalable approach to measure and compare tissue organization across spatial-omics platforms and biological scales.