Triple-negative breast cancer (TNBC) is a significant clinical challenge due to its aggressive nature and limited treatment options. In search of new treatment targets, not only single genes but also gene pairs involved in protein interactions, we explored the tumor microenvironment (TME) of TNBC from a retrospective point of view, using public single-cell RNA sequencing datasets. A High-resolution Cell type Annotation Tool, HiCAT, was used first to identify the cell type in 3-level taxonomies. Tumor cells were then identified based on the estimates of copy number variation. With the annotation results, differentially expressed genes were analyzed to find subtype-specific markers for each cell type, including tumor cells, fibroblast, and macrophage. Cell–cell interactions were also inferred for each cell type pair. Through integrative analysis, we could find unique TNBC markers not only for tumor cells but also for various TME components, including fibroblasts and macrophages. Specifically, twelve marker genes, including DSC2 and CDKN2A, were identified for TNBC tumor cells. Another key finding of our study was the interaction between the DSC2 and DSG2 genes among TNBC tumor cells, suggesting that they are more tightly aggregated with each other than those of other subtypes, including normal epithelial cells. The overexpression of DSC2 in TNBC and its prognostic power were verified by using METABRIC, a large bulk RNA-seq dataset with clinical information. These findings not only corroborate previous hypotheses but also lay the foundation for a new structural understanding of TNBC, as revealed through our single-cell analysis workflow.