Researchers from Lund University compared two powerful RNA sequencing methods, single-cell (scRNA-seq) and single-nuclei (snRNA-seq), to understand how human pancreatic islet cells express their genes. These cells, which include insulin-producing beta cells, are key to understanding diabetes and metabolic health.

While both methods analyze gene expression at the level of individual cells, only snRNA-seq can be used on frozen biobanked samples, which are often the only available tissue sources. However, most existing reference datasets were built from scRNA-seq data, raising concerns about how well they apply to snRNA-seq results.

The team analyzed samples from four human donors and compared how different annotation methods categorized cell types and detected genes. Both techniques identified similar cell types, but the proportions of those cells differed depending on the sequencing method and annotation approach. Reference-based methods were more accurate for scRNA-seq, while snRNA-seq required more careful interpretation.

Importantly, the researchers identified several new marker genes specific to snRNA-seq data, including DOCK10 and KIRREL3 for beta cells, STK32B for alpha cells, and LAMC2 for ductal cells. They also confirmed ZNF385D as a beta cell marker whose silencing reduced insulin secretion, reinforcing its role in islet biology.

This research underscores the importance of developing tailored annotation strategies for snRNA-seq data to ensure accurate cell type identification and improve our understanding of complex tissues like the pancreas.