scACCorDiON: a clustering approach for explainable patient level cell-cell communication graph analysis.

Area of Science:

• Computational Biology
• Single-cell Genomics
• Systems Biology

Background:

• Single-cell sequencing and ligand-receptor (LR) analysis are crucial for understanding cell communication in tissues.
• Directed weighted graphs model cell-cell communication, but analyzing patient cohort data remains challenging due to data variability and nonlinear communication.

Purpose of the Study:

• To develop a computational method for analyzing sample-specific cell-cell communication events from single-cell data in large patient cohorts.
• To improve the clustering of patient samples based on their disease status using cell communication profiles.

Main Methods:

• Introduced scACCorDiON (single-cell Analysis of Cell-Cell Communication in Disease clusters using Optimal transport in Directed Networks), an optimal transport algorithm.
• scACCorDiON utilizes node distances on the Markov Chain as a metric for comparing directed weighted graphs representing cell communication.

Main Results:

• scACCorDiON demonstrated superior performance in clustering disease samples compared to methods using undirected graphs.
• A case study in pancreas adenocarcinoma identified a disease sub-cluster associated with tumor microenvironment alterations.
• Detected ligand-receptor pairs were found to be predictive of pancreatic cancer survival.

Conclusions:

• scACCorDiON provides a robust and explainable method for analyzing cell-cell communication events in complex biological systems.
• The approach effectively captures disease-specific communication patterns and aids in identifying prognostic biomarkers.