Mathematically mapping the network of cells in the tumor microenvironment

Area of Science:

• Computational Biology
• Systems Biology
• Cancer Research

Background:

• Cell-cell interaction (CCI) networks are crucial for understanding disease, but current methods often generalize, missing patient-specific details.
• Existing approaches aggregate data or focus on cell types, limiting insights into individual patient heterogeneity and local network structures.

Purpose of the Study:

• To introduce RaCInG (random cell-cell interaction generator), a novel computational model for inferring personalized CCI networks.
• To leverage prior knowledge of ligand-receptor interactions and bulk RNA sequencing data for patient-specific network construction.

Main Methods:

• RaCInG utilizes a random graph-based approach to model CCI networks.
• The model integrates ligand-receptor interaction priors with bulk RNA sequencing data from 8,683 cancer patients.
• Network features were extracted to analyze the tumor microenvironment.

Main Results:

• RaCInG generated 643 network features associated with the tumor microenvironment across 8,683 cancer patients.
• These features showed significant associations with immune response and cancer subtypes.
• The model enabled the prediction and explanation of immunotherapy responses, demonstrating robustness and consistency with existing methods.

Conclusions:

• RaCInG provides a powerful tool for elucidating patient-specific network dynamics in cancer.
• The findings offer novel insights into cancer biology and personalized treatment responses.
• RaCInG has the potential to advance the study of complex CCIs in cancer and other biomedical fields.