Single-cell transcriptomic reveals network topology changes of cancer at the individual level

Abstract

Network biology facilitates a better understanding of complex diseases. Single-sample networks retain individual information and have the potential to distinguish disease status. Previous studies mainly used bulk RNA sequencing data to construct single-sample networks, but different cell types in the tissue microenvironment perform significantly different functions. In this study, we investigated whether network topology features of cell-type-specific networks varied in different pathological states at the individual level. Protein-protein interaction network (PPI) and co-expression network of cancer and ulcerative colitis were established using four publicly single-cell RNA sequencing (scRNA-seq) datasets. We analyzed cell-cell interactions of epithelial cells and immune cells using CellChat R package. Network topology changes between normal tissues and pathological tissues were analyzed using Cytoscape software and QUACN R package. Results showed cell-cell interactions of epithelial cells were enhanced in carcinoma and adenoma. The average number of neighbors and graphindex of co-expression network increased in epithelial cells of adenoma, carcinoma and paracancer compared with normal tissues. The co-expression network density of T cells in tumors was significantly higher than that in normal tissues. The co-expression network complexity of epithelial cells in the benign tissues was associated with the grade group of paired tumors. This study suggests topological properties of cell-type-specific individual network vary in different pathological states, providing an insight into understanding complex diseases.