Now,new research fromtheStowers Institute for Medical ResearchandHemholtz Munich, published May 11, 2026, inCell,hasdeveloped a new AI framework designed to help answer that question. RegVelo is a model that connects two areas of single-cell biology that have oftenremainedseparate: methods that estimate how cells change over time and methods that infer the gene regulatory networks controlling those changes.
By bringing those pieces together,RegVeloallows researcherstotime travel, predict how cells change, and identify whichgenes control those changes. They can do itall throughcomputer simulations, eliminating the needto run every experiment in the lab.
“So, whyisthisimportant toknow?”saidTatjana Sauka-Spengler, Ph.D.,Stowers Institute Investigatorand co-senior author of the study. “Youcanimagine, if you hada very earlyset of cells,being equipped with a particular set of instructions couldallow you to reproduce, in vitro, some of these celltypesin a very natural way. These cells could then be used in cell therapiesin regenerative medicine.”
Video: Predicting cell fate, hear from the scientists behind the work
“Sauka-Spengler and her collaborators have developed a meaningfully different way to process this kind of data,”said Stowers President and Chief Scientific OfficerAlejandro Sánchez Alvarado, Ph.D.”Itallows us to infer the most likely path of each component through space and time, and to use deep learning to predict thosedynamicsand test them experimentally.”
Inthestudy,RegVelomodeledthe neural crest,a group of early embryonic cells that can become manydifferent partsof the body.In zebrafish neural crest development, RegVeloidentifiedan early driver of pigment cell formation (tfec) and revealed a previously unknown regulator of pigment cell fate (elf1). Those predictions were then supported experimentally, showing that the model could do more than describe developmental change.
“There is always an initiating, driving element in something that will be defined at the end,” said Sauka-Spengler. “But most times,if not always, that element is lost if you’re onlyanalyzingthe final cell state. Development is often described as a series of static snapshots of cell states. What we really want to understand, however, is how cells make decisions—how they transition from one state to another.RegVelomodels how these fate decisions are encoded in gene regulatory networks over time and space, and what drives them.”
By helping connect early regulatory events to later cell fates, the work could also improve howscientistsstudy developmental disorders and, over time, help guide efforts in regenerative medicine and cell therapy.
“RegVelo’svalue extends well beyondneural crest cells,” said Sánchez Alvarado. “It’sapplicable to any system in which cells change over time, from basic developmental biology to modelingtumortrajectories and the cellular outcomes that may inform treatment. It deserves attention from anyone working on cellular dynamics.”
Bridging a long-standing gap in single-cell biology
Single-cell biologyresearchhas made it possible to build increasingly detailed maps of development. RNA velocity methods can help researchers estimate how cells move through developmental landscapes, while gene regulatory network approaches canidentifyrelationships among genes. But those methods have typically been used in parallel rather than together. RNA velocity methods often do not directly model transcriptional regulation, while regulatory network approachesgenerally donot capture cellular dynamics over time.
“For a long time, cellular dynamics and gene regulation have largely been modeled separately,” said the study’s co-senior author Prof. Fabian J. Theis, Ph.D., Director of the Computational Health Center (CHC) at Helmholtz Munich and Professor at the Technical University of Munich (TUM). “RegVelobrings those pieces together, allowing us to ask not only how cells are changing, but which regulatory interactions are helping drive those changes.”
The framework jointly models splicing kinetics and gene regulatory relationships, allowing researchers tomap the hidden timeline of cell development, predicthow cells shift from one state to another, and test what might happenwhen specific regulators are perturbed. In practical terms, that means scientists can ask a more mechanistic question than simply”Where is this cell going?”Now they can ask,”Which genes are helping push it there?”
Joining forces
The work also reflects a deep collaboration between complementary teams.Sauka-Spengler’sLab,which transitioned from theUniversity of Oxfordto the Stowers Institute in 2022,broughta high-resolution gene regulatory scaffold for cranial neural crest development, while Theis’s group broughtcomputationalexpertiseinmodelingand definingdevelopmental trajectoriesof single cells andusing RNAvelocityanalyses. Together, those approaches were combined into a shared deep learning modelthat made developmental transitions highly predictive and testable.
“What made this work especially powerful was the combination of complementary strengths…high-resolution gene regulatory circuitry fromour laband dynamic trajectoryand networkmodelingfromFabian’s team,” Sauka-Spenglersaid. “RegVelo emerged from integrating those two views into one framework.”
What are Gene Regulatory Networks?
Video: Tatjana Sauka-Spengler, Ph.D., on decoding the cell’s instructions
Generegulatory networks are the ordered sets of instructions that help guide a cell from one identity to another.
“They area cascade of events,”Sauka-Spengler explained.”Onegroup of genes activates or suppresses another, pushing a cell down one path instead of another. That matters because every cell in the body starts with the same DNA. What makes a skin cell different from a neuron or a muscle cell is not the genome itself, but which genes are turned on, when they are activated, and in what combinations.”
The process can be compared to an electronic circuit.Some genes act like “go” signals, others like brakes, and together they create a code that scientists are trying to decipher.
How it works: a deeper dive into the findingsand implications
TheteamusedRegVelotoidentifyan early pigment driver and revealed a previously unknown regulator of pigment cell fatein zebrafish. The researchers applied the frameworkacross multiple systems, including cell cycle, pancreatic endocrinogenesis,hematopoiesis, myogenesis, hindbrain development, and zebrafish neural crest development. Across these settings, the model performed comparably to or better than leading approaches in inferring latent time, velocity, terminal states, and lineage-associated drivers.
One of the most compelling examples came in the neural crest, a developmental system that gives rise to many different cell types, including pigment cells, craniofacial tissues, and parts of the peripheral nervous system. There,RegVeloproved especially useful because it couldidentifyregulators acting early in a developmental trajectory,even when those genes were not strongly expressed in the final cell state.
Using that approach, the researchers found thattfec appears to act as an early driver of pigment cell development. They also identified elf1 as a previously unknown regulator of pigment lineage fate.
Follow-up experiments, including CRISPR/Cas9-mediated knockout and single-cell Perturb-seq, supported both predictions, showing that the model could do more than describe developmental change—it could generate biologically meaningful hypotheses that held up in living systems.
“RegVelois a model that integrates these two knowledges and basically allows us to prove the validity of our discoveries,”Sauka-Spengler said. “We’re able to predictthe essential drivers of particular cell fate and cell commitments, and then we’re able to simulate this perturbation and read out clearly the effect on very downstream outcomes.”
That capability matters, in part, because of scale. Researchers are often dealing with hundreds, andsometimes thousands, ofpossible factors. Experimentally perturbing each of them one by one would be costly and impractical.RegVelohelps narrow that search.
“Because we’re talking about networks involving hundreds, and sometimes multiple hundreds of genes, imagining that we would want to perturb all of them andanalyzeall of them would not really be feasible,”Sauka-Spengler said. “So,we can useRegVeloas both analysis and a prediction and screening tool for future experiments.”
Computationalmodelingwith experimental validation: a more predictive and promising combination
The teamdescribesRegVeloas a step toward a more predictive form of developmental biology,in which computational models help prioritize experiments, uncover hidden regulators, and forecast how cell fates may change when gene networks are perturbed.RegVeloreveals how cells transition between states andidentifiesthe gene interactions that drive their fate decisions.
The implications could extend well beyond one developmental system. High-resolution regulatory understanding could help researchers betteridentifythe causes of developmental defects and, over time, direct cells more precisely incell therapies andregenerative medicine— from repairing heart muscleand growing skin grafts to developing lab-growncartilage.Craniofacial disorders, pigment cell defects, and broader efforts to guide stem cells or organoids toward desired cell states are among the areas where that deeper understanding could matter.
“Having a full resolution of gene regulatory circuitry that has been predicted, simulated, perturbed and then validated gives us a very solid tool in our hands,”Sauka-Spengler said. “We can start from stem cells or from naive cells and develop new ways of directing them towards the cell types that can then be used in cell therapies for purposes of treatments.”
Thestudy suggeststhat the framework could be extended in the future to incorporate additionalregulatory layers, including chromatin, protein activity, and other multimodal measurements.
While there arecurrent limitations —including simplifying assumptions around latent time, regulatory interactions, and computational cost — the study offers a compelling proof of principle. “When dynamic cell-statemodelingis linked directly to gene regulation, it becomes possible to move closer to mechanism and then discovery,” Sauka-Spengler explained.
More information:
Read the press release from Helmholtz Munich here.
Additionalauthors include Weixu Wang, Zhiiyuan Hu, Philipp Weiler, Sarah Mayes, Marius Lange, Daniel M. Fountain, Julianna O. Haug, Jingye Wang, and Zhengyuan Xue
Research at the Stowers Institute was supported by institutional funding to Tatjana Sauka-Spengler, Ph.D., and byWellcomeTrust Award 215615/Z/19/Z. Additional support for the broader collaborative work came from the European Union/ERCDeepCellproject (101054957), theWellcomeLeapΔTissueProgram (9E8E84F7-8991-4D4A-A9EC), the European Union’s Horizon 2022 research and innovation programme (101057775), the German Federal Ministry of Education and Research through the HOPARL project (031L0289A), the DFG Graduate School of QBM (GSC 1006), the Joachim Herz Foundation, an EMBO Postdoctoral Fellowship, the Fundamental Research Funds for the Central Universities (2042025kf0022, 2042022dx0003), and the National Natural Science Foundation of China (32500725).
About the Stowers Institute for Medical Research
Founded in 1994 through the generosity of Jim Stowers, founder of American Century Investments, and his wife, Virginia, the Stowers Institute for Medical Research is a non-profit, biomedical research organization with a focus on foundational research. Its mission is to expand our understanding of the secrets of life and improve life’s quality through innovative approaches to the causes, treatment, and prevention of diseases.
The Institute consists of 24 independent research programs. Of the approximately 500 members, over 370 are scientific staff that include principal investigators, technology center directors, postdoctoral scientists, graduate students, and technical support staff. Learn more about the Institute at www.stowers.org and about its graduate program at www.stowers.org/gradschool.
Media Contact:
Joe Chiodo, Director of Communications
724.462.8529
[emailprotected]
SOURCE Stowers Institute for Medical Research