Inferring Personalized Cell-Cell Communication Networks in Colorectal Cancer with Individualized Causal Discovery.

Area of Science:

• Computational biology
• Precision oncology
• Cancer research

Background:

• Tumor heterogeneity poses challenges for understanding cell-cell communication networks (CCCNs).
• Existing methods fail to capture patient-specific signaling patterns due to reliance on population-level data.
• Individualized CCCNs are crucial for precision oncology.

Purpose of the Study:

• To develop an integrative computational framework for inferring individualized CCCNs (iCCCNs).
• To identify hierarchically structured gene expression modules (GEMs) and patient-specific signaling patterns in colorectal cancer (CRC).
• To validate the clinical relevance of iCCCNs as prognostic signatures.

Main Methods:

• Combined nested hierarchical Dirichlet process (nHDP) for GEM identification with instance-specific Greedy Fast Causal Inference (iGFCI) for iCCCN inference.
• Applied the framework to single-cell RNA-seq data from over 625,000 cells.
• Utilized TCGA bulk RNA-seq data and survival data for validation.

Main Results:

• Successfully decomposed complex GEMs and uncovered iCCCNs across detailed cell subtypes in CRC.
• Validated the clinical relevance of individualized GEM causal interactions as prognostic signatures.
• Identified ligand-receptor pairs mediating cell-cell communication and enabling insights into immune evasion.

Conclusions:

• The developed computational framework advances personalized oncology by enabling precise patient stratification.
• Identified iCCCNs and GEMs offer actionable biomarkers for improved therapeutic targeting in cancer.
• This approach provides mechanistic insights into immune evasion and personalized cancer treatment.