CSsingle: A Unified Tool for Robust Decomposition of Bulk and Spatial Transcriptomic Data Across Diverse Single-Cell...

Area of Science:

• Genomics
• Bioinformatics
• Cancer Research

Background:

• Cellular heterogeneity poses challenges for transcriptomic data analysis.
• Accurate deconvolution of bulk and spatial transcriptomic (ST) data is crucial for understanding tissue architecture and disease.
• Existing methods struggle to account for variations in RNA content between cell types.

Purpose of the Study:

• To introduce CSsingle, a novel computational method for enhanced decomposition of bulk and ST data.
• To provide a unified tool for integrating single-cell, bulk, and ST data.
• To identify cellular biomarkers for predicting treatment outcomes in esophageal adenocarcinoma.

Main Methods:

• CSsingle applies cell size correction using ERCC spike-in controls for accurate bulk data deconvolution.
• The method enables fine-scale analysis of ST data to study tissue microenvironments.
• Benchmarking against existing methods was performed to assess accuracy and robustness.

Main Results:

• CSsingle outperforms current methods in accuracy and robustness.
• Validation in over 700 gastroesophageal samples identified mosaic columnar cells (MCCs) predominantly in Barrett's esophagus and esophageal adenocarcinoma (EAC).
• A dynamic relationship between MCCs and squamous cells during immune checkpoint inhibitor (ICI) treatment in EAC patients was revealed.

Conclusions:

• MCCs play a critical role in EAC treatment and are potential biomarkers for predicting immunochemotherapy outcomes.
• MCC expression signatures can guide personalized immunotherapeutic strategies for EAC.
• CSsingle advances the study of complex biological systems and disease processes by integrating multi-omic data.