Comparison and optimization of cellular neighbor preference methods for quantitative tissue analysis.

Area of Science:

• Spatial biology
• Computational pathology
• Bioinformatics

Background:

• Cellular spatial distribution is crucial for tissue function in health and disease.
• Pairwise neighbor preference analysis quantifies cell type co-occurrence in spatial omics.
• Existing methods lack clear guidance for selecting the optimal approach.

Purpose of the Study:

• To comprehensively compare existing neighbor preference analysis methods.
• To introduce and evaluate a novel method, COZI (Combination of Analysis Steps).
• To assess methods' ability to distinguish tissue architectures and recover neighbor preference directionality.

Main Methods:

• Systematic evaluation of analytical steps in existing neighbor preference methods.
• Introduction of COZI, a novel combination of analysis steps.
• Comparison of methods using tissue simulations and biological datasets.

Main Results:

• Delineation of method-specific strengths and limitations in spatial omics analysis.
• Demonstration of COZI's superior sensitivity and directionality in neighbor preference analysis.
• Validation of COZI's performance on simulated and real-world tissue data.

Conclusions:

• COZI offers a sensitive and directional approach for quantifying spatial cell type interactions.
• The study provides guidance for selecting appropriate neighbor preference analysis methods.
• Accurate spatial analysis is vital for advancing our understanding of tissue function and disease.