Crosstalk In: crosstalk-aware inference of tumor microenvironment cell infiltration

Area of Science:

• Computational Biology
• Cancer Research
• Bioinformatics

Background:

• The tumor microenvironment (TME) is crucial for cancer progression and treatment outcomes.
• Current bulk deconvolution methods often overlook cell-cell interactions within the TME.

Purpose of the Study:

• To develop a crosstalk-aware deconvolution framework for TME analysis.
• To improve the accuracy of cell-type composition inference and its clinical applications.

Main Methods:

• Developed CrosstalkIn, a framework integrating Gene Ontology and protein-protein interaction data.
• Constructed patient-specific cell-cell networks and applied a modified random walk with restart algorithm to calculate cell Infiltration Scores (InScores).

Main Results:

• CrosstalkIn demonstrated superior deconvolution performance across validated cohorts.
• In lower-grade glioma, InScores identified survival-associated cell types and stratified patients.
• In adenocarcinoma and melanoma, CrosstalkIn showed prognostic value and improved immunotherapy response prediction.

Conclusions:

• CrosstalkIn offers a robust method for inferring TME cell infiltration from bulk transcriptomic data.
• The framework enhances cancer prognosis and immunotherapy response prediction.