scE2TM improves single-cell embedding interpretability and reveals cellular perturbation signatures

Single-cell RNA sequencing reveals cellular heterogeneity, yet computational methods struggle to balance performance with biological interpretability. Embedded topic models provide interpretable cell representations, but may learn overly similar topics, resulting in redundancy and incomplete capture of biological variation. Single-cell foundation models create opportunities to harness external biological knowledge for guiding model embeddings. Here, we present scE2TM, an external knowledge-guided embedded topic model for interpretable scRNA-seq analysis. scE2TM implements embedding clustering regularization where each topic is encouraged to represent a distinct group of genes, enabling it to capture unique biological information. We show that across 20 datasets, scE2TM outperforms seven state-of-the-art methods in clustering performance. We perform an interpretability benchmark to show that scE2TM topics exhibit greater diversity and stronger consistency with biological pathways. When modelling interferon-stimulated peripheral blood mononuclear cells, we find that scE2TM simulates topic perturbations that shift control cells toward stimulated states, recapitulating experimental interferon responses. When tested on a melanoma dataset, scE2TM identifies malignant-specific topics and extrapolates them to unseen patient data, highlighting melanoma-associated gene programs linked to patient survival. Computational methods for scRNA-seq data struggle to balance performance with biological interpretability. Here the authors develop scE2TM, which improves data clustering, while learning biologically meaningful and interpretable gene programs, revealing insights into disease and patient prognosis.