New software simulates how cancer cells grow and spread

What if we could simulate cancer like a storm?

Researchers at the University of Maryland School of Medicine (UMSOM) have developed new software that can predict growth, spread, and interactions of cancer cells by combining actual patient genomic data with mathematical modeling. The program uses purely English “grammar” to describe cellular behavior, making it easier for interdisciplinary scientists to build virtual tissues and predict disease progression. This could lay the foundation for digital twins and fully virtual clinical trials for cancer patients.

A language that simulates life

The study was published on July 25 cella novel approach is introduced that translates biological processes into readable code, using a “hypothetical syntax” that researchers call. Scientists can now write cancer models using software to convert common language statements of cellular behavior. This is a bit like scripting simulation, but for biological organization.

“While this new ‘grammar’ can communicate between biology and code, it also enables communication between scientists to leverage this modeling paradigm in their research,” said Daniel Bergman of the UMSOM Institute of Genomic Sciences (IGS).

From snapshot to simulation

Traditional genomic tools provide only static snapshots of units. But cancer is a dynamically developed system. New software models cancer is produced by complex cellular communication and how it responds to treatment over time.

“Cancer is controlled or implemented by a highly personalized immune system; this complexity makes it difficult to predict specific patients from human cancer data,” said Dr. Jeanette Johnson, a first and postdoctoral researcher at IGS. “This framework provides us with a sandbox to freely study our hypotheses…no additional costs or risks.”

What the model reveals

To test their system, the team combined its modeling language with spatial transcriptomic data from real human tissue. They simulated:

• The immune system cannot control breast tumor attack
• Pancreatic cancer response to immunotherapy using untreated patient tissue samples
• Communication between fibroblasts and tumor cells in pancreatic tumors

Each digital patient in the simulation responds to treatment differently, emphasizing the diversity and complexity of the tumor microenvironment. These “virtual patients” may one day help tailor the treatment plan without having to test the actual patient first.

From cancer to brain

To demonstrate the broader value of grammar, researchers at Johns Hopkins used the software in neuroscience experiments to model the development of the brain layer. This flexibility suggests that the method can be applied in many areas of biomedical research.

“We have a new biological research framework because researchers can now perform computerized simulations of their benchtop experiments and clinical trials,” said UMSOM President Mark T. Gladwin. “This has important applications to conduct digital twins and virtual clinical trials in cancer and beyond.”

Powered by collaborative and open source tools

The open source of syntax and modeling systems is key. “By giving the scientific community access to this tool, we are providing a pathway to standardize this model,” Dr. Bergman said.

Senior writer Dr. Elana J. Fertig said the work was a “tapestry of team science.” She sees similarities between predictive weather systems and predictive biological systems. “Adapting this approach to genomics provides us with a virtual cellular laboratory where we can experiment … entirely in a computer.”

The project received funding from the National Cancer Research Foundation, as well as several grants from the National Cancer Institute.

Journal Reference

Magazine: cell
Publication date: July 25, 2025
title: Human-interpreted syntax encodes biological models of multicellular systems to democratize virtual cell laboratories

Related