Pancreatic cancer (PAC) remains one of the most lethal malignant neoplasms, which is diagnosed at an advanced stage and thus lose the chance for curative resection. Here, we further probed PAC with a comprehensive multi-omics approach. Using single-cell RNA sequencing, we provided an integrated analysis of ductal cell subpopulations over the Leiden algorithm to identify two mian subcluster: S100A6 + cells and FXYD2 + cells. The gene set enrichment analysis results show that the two subtypes focused on different pathways related to tumor development. Furthermore, we integrated bulk and single-cell RNA sequencing datasets to generate and validate the prognostic signatures of the overall survival (OS) in PAC patients and S100A6 + cells were significantly enriched in high-risk groups which had a poor prognosis. Collectively, this research expands our understanding of ductal cell and provides a new reliable prognosis signature in PAC.