Researchers from Uniklinik RWTH Aachen, in collaboration with RWTH Aachen University, have developed a new computational method, PHLOWER, that can decipher complex cell differentiation processes from single-cell multi-omics data. This novel method will transform how researchers study organ development, regenerative medicine, and disease mechanisms.
Cell differentiation — the process by which stem cells transform into specialized cell types — lies at the heart of organ formation and disease progression. While single-cell sequencing technologies can capture snapshots of gene activity and chromatin accessibility across thousands of individual cells, reconstructing the dynamic “trajectories” of how these cells evolve has remained a major computational challenge.
PHLOWER (decomPosition of the Hodge Laplacian for inferring trajectOries from floWs of cEll diffeRentiation) addresses this challenge by leveraging mathematical tools from topology — specifically the discrete Hodge decomposition — to map how cells transition through complex, multi-branching differentiation pathways. Unlike previous trajectory inference methods that struggle with large or intricate lineage trees, PHLOWER can robustly infer high-dimensional trajectories from both RNA and chromatin data.
“PHLOWER introduces a fundamentally new way of representing single cell data under a differentiation process,” said Prof. Ivan G. Costa, senior author from Uniklinik RWTH Aachen. “By representing cellular transitions as flows, we can identify subtle branching events and uncover the regulatory factors that guide them.”
A New Lens on Kidney Organoid Development
The researchers applied PHLOWER to study human kidney organoids, which are 3D mini-organs derived from induced pluripotent stem cells (iPSCs). These organoids are invaluable tools for understanding kidney development, modeling diseases, and testing drugs. However, their production often suffers from “off-target” differentiation, where cells deviate from kidney lineages and form unwanted cell types such as neuronal or muscle cells. These off-target populations can compromise organoid function and limit their use in research and therapy.
By analyzing multimodal single-cell data from kidney organoids at multiple stages of differentiation, PHLOWER reconstructed a detailed cellular trajectory map showing how progenitor cells branch into specific kidney lineages — such as podocytes and tubular cells — as well as off-target neuronal and stromal fates.
“The multimodal single-cell protocol was crucial for this study,” said Univ.-Prof. Dr. med. Christoph Kuppe, Senior Physician at the Department of Nephrology and Hypertension, Rheumatology and Immunology (Department of Medicine II) at Uniklinik RWTH Aachen. “Not only could we measure gene activity and chromatin accessibility simultaneously, but we could map out the spatial localization and interaction using single-cell resolved spatial transcriptomics. This gave us a comprehensive picture of how differentiation is orchestrated over time and space — something that would be impossible to achieve with a single-data modality alone.”
Crucially, PHLOWER identified key transcription factors (TFs) that drive these divergent pathways. Among them were PAX3, RFX4, and ZIC2, which the algorithm predicted to promote unwanted neuronal differentiation. When the team experimentally silenced these genes in developing kidney organoids, they observed a striking reduction in off-target cells and a boost in the formation of kidney-specific cell types, including podocytes and tubule epithelial cells.
“Kidney organoids are a powerful system to model development and disease, but their reproducibility has been limited by off-target differentiation,” explained Univ.-Prof. Dr. med. Rafael Kramann, Director of the Department of Nephrology and Hypertension, Rheumatology and Immunology (Department of Medicine II) at Uniklinik RWTH Aachen. “PHLOWER provides a roadmap to identify and correct these unwanted trajectories, bringing us closer to generating more mature, functional kidney tissues in the lab.”
PHLOWER showcases the interdisciplinary collaboration across computational and experimental sciences at RWTH Aachen promoted via the Center of Computational Life Sciences RWTH.
“This work exemplifies the mission of the RWTH Center for Computational Life Sciences (CCLS),” said Prof. Michael T. Schaub, co-author of the study and Professor for Computational Network Science, a new professorship bridging Computer Science and Biology at RWTH Aachen University. “At CCLS, we unite expertise from medicine, biology, mathematics, and computer science to drive data-driven innovation in the life sciences. By applying cutting-edge computational modeling and AI to biological data, we aim to transform complex datasets into actionable insights that advance health, sustainability, and biotechnology.”
The work was supported by the German Research Foundation (DFG), European Research Council (ERC), and the German Federal Ministry of Education and Research (BMBF).
Cheng, M., Jansen, J., Reimer, K.C. et al. PHLOWER leverages single-cell multimodal data to infer complex, multi-branching cell differentiation trajectories. Nat Methods (2025). doi.org/10.1038/s41592-025-02870-5
Contact
Prof. Ivan G. Costa
Institute for Computational Genomics
Uniklinik RWTH Aachen
ivan.costa@rwth-aachen.de
Prof. Rafael Kramann
Institute for Experimental Internal Medicine and Systems Biology
Uniklinik RWTH Aachen
rkramann@ukaachen.de
