Comparison of methods and resources for cell-cell communication inference from single-cell RNA-Seq data

Area of Science:

• Computational biology
• Single-cell genomics
• Systems biology

Background:

• Single-cell transcriptomics data is rapidly expanding, driving interest in inferring cell-cell communication.
• Numerous computational tools exist, each using specific interaction databases and prediction methods.
• The influence of resource and method selection on cell-cell communication predictions remains unclear.

Purpose of the Study:

• To systematically compare existing cell-cell communication inference resources and methods.
• To evaluate the impact of resource and method choices on prediction outcomes.
• To assess the coherence of predictions with spatial, cytokine, and protein abundance data.

Main Methods:

• Systematic comparison of 16 cell-cell communication inference resources and 7 prediction methods.
• Analysis of resource overlap, pathway coverage, and protein enrichment.
• Evaluation of method-resource combinations and agreement with spatial colocalization, cytokine activity, and receptor abundance data.
• Development of the LIANA (LIgand-receptor ANalysis frAmework) tool.

Main Results:

• Resources exhibit limited unique interactions, variable overlap, and uneven pathway/protein coverage.
• Both methods and resources significantly influence predicted intercellular interactions.
• Cell-cell communication predictions generally align with spatial colocalization, cytokine activity, and receptor protein abundance data.

Conclusions:

• The choice of resource and method is critical in cell-cell communication inference.
• Predictions show general coherence with multiple biological data types.
• LIANA provides an open-source framework for standardized analysis of cell-cell communication.