Identifying condition-related cell-cell communication events using supervised tensor analysis.

Area of Science:

• Computational Biology
• Genomics
• Immunology

Background:

• Cell-cell communication (CCC) is vital for biological processes and disease pathogenesis.
• Current methods for inferring CCC from single-cell RNA sequencing (scRNA-seq) data often overlook crucial confounding factors like batch and demographic variables.
• Analyzing multi-sample, multi-condition scRNA-seq data requires robust methods that can handle complex variations.

Purpose of the Study:

• To introduce STACCato, a novel supervised tensor analysis tool for cell-cell communication (CCC) inference.
• To develop a method that identifies CCC events and quantifies the impact of biological conditions while adjusting for confounders.
• To provide a computational framework for more accurate CCC analysis in complex scRNA-seq datasets.

Main Methods:

• Developed STACCato, a supervised tensor analysis framework for CCC inference.
• Integrated adjustment for sample-level confounders (e.g., batch, demographics) into the CCC analysis pipeline.
• Applied STACCato to simulated datasets and real scRNA-seq data from lupus and autism studies.

Main Results:

• STACCato accurately identifies cell-cell communication events and their condition-specific effects.
• The method demonstrated superior performance in estimating disease effects compared to tools ignoring sample-level variables.
• Analysis of lupus and autism data revealed more precise cell type activity patterns using STACCato.

Conclusions:

• STACCato offers a robust approach for CCC inference in scRNA-seq data by accounting for confounders.
• Incorporating sample-level variables significantly enhances the accuracy of disease effect estimation and cell type activity profiling.
• The STACCato framework provides a valuable tool for advancing our understanding of CCC in health and disease.