De Novo Detection of Clonal Structure and Evolution in Single-Cell and Spatial Transcriptomes

Tumors are composed of cellular populations with distinct genotypes and phenotypes, which dynamically evolve over time and during treatment. This process is known as clonal evolution, and it is difficult to reveal fine-scale clonal structure with traditional bulk sequencing. Although single-cell genome sequencing could enable reconstruction of tumor clonal evolution, it remains technically challenging and the number of single cells profiled is generally insufficient due to high cost. To address this issue, we developed scClone, a computational toolkit that integrates variant detection and genotype inference for single-cell RNA-seq (scRNA-seq) and spatial transcriptomic data. It further provides interactive visualization of clonal structure and dynamic evolution. scClone addresses key limitations inherent to scRNA-seq, such as expression drop-out and allelic imbalance, and incorporates cell type or state annotation with mutational signature analysis to enable comprehensive profiling of tumor heterogeneity. scClone demonstrated robust performance across multiple datasets-generated from both full-length and fragmented RNA sequencing-by accurately reproducing mutation profiles and resolving clonal mixtures in myeloma, hepatocellular carcinoma and pancreatic cancer. Additionally, scClone has been applied to spatial transcriptomics, enabling the delineation of clonal structures within histological sections from ovarian cancer and cutaneous squamous cell carcinoma. In summary, our results demonstrate that scClone can extract genetic information from scRNA-seq datasets, thereby successfully establishing genotype-phenotype associations at the single-cell level and providing insights into the clonal evolution of tumors.