A set of 67 gene markers is giving stem cell researchers a clearer way to tell where developing cells are headed. Built by scientists in Dusseldorf, the panel separates cells that have not yet begun to specialize from cells moving toward three major developmental lineages.

The work addresses a challenge in studying human pluripotent stem cells, or hPSCs: cells that can develop into many types of tissue. As they change, they pass through short-lived, partly overlapping stages. A shift in gene activity may signal a temporary transition, not a firm commitment to become a particular kind of cell. Existing markers do not always make that distinction reliably, which can complicate experiments and the interpretation of mixed cell samples.

Researchers Jochen Dobner, Alessandro Prigione and Andrea Rossi, from the IUF—Leibniz Research Institute for Environmental Medicine and Heinrich Heine University Duesseldorf, assembled a carefully tested panel to help. It includes 20 markers for undifferentiated cells, 17 for endoderm, and 15 each for ectoderm and mesoderm. These three groups, known as germ layers, give rise to different tissues in the body.

The team began with 172 candidate genes identified through RNA sequencing, a method that measures which genes are active. They tested the candidates using qPCR, a technique for measuring genetic material, in several independent stem cell lines. They also combined the results with markers from their earlier work.

The panel could sort publicly available RNA-sequencing data into undifferentiated cells and the three germ layers. Its marker patterns also matched the corresponding cell groups in reference data from human and macaque embryos during gastrulation, an early stage of development when these lineages take shape. For some markers, the researchers checked the results at the protein level in the KOLF2.1J reference cell line, using broad protein profiling.

The team has also made the findings easier to use through DeepDiff, a free web application. It displays gene activity across 33 early hPSC differentiation states and includes protein data for KOLF2.1J. The site also offers an analysis comparing gene and protein levels, plus a machine-learning classifier that can automatically label bulk RNA-sequencing data. Bulk data combine signals from many cells, so identifying the cell states within them can be difficult.

Together, the marker panel and DeepDiff give researchers tools to assess stem cell cultures more consistently, including samples that contain a mix of cell states. That could help make studies easier to compare and clarify whether a culture is still undifferentiated or moving toward a specific lineage. The research was published in Stem Cell Reports in 2026 as “Defining early human developmental identity: A curated, cross-platform marker framework” (DOI: 10.1016/j.stemcr.2026.103123). With the web tool freely available, labs can now explore the data and classifier as they study the earliest steps toward specialized cells.