UBS93 enables delineating intratumor subtype heterogeneity in breast cancer

Breast cancer is a heterogeneous disease that comprises multiple molecular subtypes. Although several methods have been developed to subtype bulk tumor samples, their capacity to delineate intratumor heterogeneity at the single-cell level remains limited. In this study, we present UBS93, a novel classifier specifically designed for breast cancer subtyping in both bulk tumor samples and individual cancer cells. We rigorously validate UBS93 and demonstrate its superior performance compared to existing methods, particularly in the identification of Claudin-low samples. Application of UBS93 to single-cell RNA sequencing (scRNA-seq) data from triple-negative breast cancer patients uncovered a previously unrecognized "Basal-like/Claudin-low mixed" subtype, suggesting a Basal-like origin for Claudin-low cancer cells. Subsequent computational analyses and experimental validations reveal that downregulation of transcription factor ELF3 induces a partial Claudin-low phenotype in Basal-like breast cancer cells. Collectively, our findings provide a powerful tool for elucidating intratumor subtype heterogeneity in breast cancer, with potential applicability to other cancer types.