PUMC_CHINA

Abstract

The single-cell RNA sequencing (scRNA-seq) technology provides a powerful tool for dissecting heterogenous tumor tissue at single cell resolution. However, it still remains challenging to reconstruct cell differentiation dynamics on the expression spectrum. Here, we develop a new method combining lineage clustering and generalized additive model (GAM). Validating this method on hematopoietic cell lineage and real hepatocellular carcinoma (HCC) samples, we find our method have a greater potential to build differentiation pseudotime with a higher resolution. We also find that enrichment of specific lineage and differential expression patterns are associated with worse clinical outcomes. Combining theses together, we believe our method could provide better a tool for studying heterogenous biopsies such as tumor.