Scientists advance efforts to create ‘virtual cell lab’ as testing ground for future research with live cells

Using mathematical analysis of patterns of human and animal cell behavior, scientists say they have developed a computer program that mimics the behavior of such cells in any part of the body. Led by investigators at Indiana University, Johns Hopkins Medicine, the University of Maryland School of Medicine and Oregon Health & Science University, the new work was designed to advance ways of testing and predicting biological processes, drug responses and other cell dynamics before undertaking more costly experiments with live cells.

With further work on the program, the researchers say it could eventually serve as a “digital twin” for testing any drug’s effect on cancer or other conditions, gene environment interactions during brain development, or any number of dynamic cellular molecular processes in people where such studies are not possible.

The new study and examples of cell simulations are described online July 25 in the journal Cell.

According to Genevieve Stein-O’Brien, Ph.D., the Terkowitz Family Rising Professor of Neuroscience and Neurology at the Johns Hopkins University School of Medicine, the research project began at a workshop for an earlier version of computer software, called PhysiCell, designed by Indiana University engineering professor Paul Macklin, Ph.D.

PhysiCell is based on so-called agents, “essentially, math robots that act on [a set of] rules that reflect the cells’ DNA and RNA,” says Stein-O’Brien. Each type of cell in the body is mapped to an agent and then digitally manipulated to do things, such as interact with other cells and environmental factors such as therapeutics, oxygen, and other molecules in the process of form tissues, organs and sometimes, cancer.

By tracking cells following their assigned rules, scientists can virtually see such things as how tumors emerge and interact with therapeutics and the immune system. They can track cells that form layers of the brain’s cortex, and see how brain cells organize to lay the foundation they will need to create circuits.

Stein-O’Brien’s lab, in collaboration with co-first author Daniel Bergman, Ph.D., assistant professor at the University of Maryland School of Medicine’s Institute for Genome Sciences, is leading the further development of the software to go all the way from cells to circuits in the brain.

Macklin says that typical computer modeling programs exist but generally require sophisticated knowledge of math models and computer coding to use and interpret. The new PhysiCell software, he says, formulated a new “grammar” that makes the agent-based computer model more accessible to scientists who know a lot about biology but aren’t proficient in programming.

“It used to take months to write the code for these models, and now we can teach other scientists to create a basic immunology model in an hour or two,” says Macklin. “We can also use this program to model spatial transcriptomics, a longtime goal for scientists, to visualize where each cell type can be found and how they function in 3D replicas of tissues and tumors.”

Stein-O’Brien describes the new coding grammar as “literally, an Excel spreadsheet that, on each line, matches a cell type with a rule in human legible syntax. For example: this cell increases division as oxygen concentration increases.”

Then, the program automatically translates the biological grammar from the spreadsheet into math equations that produce a guide for cell behavior. The program can also tune the model to match established data from studies of the transcriptome, the output of genetic material.

Study author David Zhou, a Johns Hopkins University Neuroscience undergraduate student at the time, worked with Stein-O’Brien to provide many of the cell behaviors included in the new program. He and Zachary Nicholas, a Johns Hopkins Human Genetics Ph.D. candidate and NIH/NINDS D-SPAN Scholar, built the model of brain development—believed to be the first of its kind—using data from the Allen Brain Atlas.

This was enabled by new advancements in software that uses spatially resolved data to connect snapshots of cell behavior to build a movie that shows cell and tissue interactions over time. “This is very important for human disease,” says Stein-O’Brien. “We want to test changes in the cell rules, patterns and paths to see how cells change their behavior.”

The models involving cancer cell behavior were initially based on data from a large collection of human pancreatic tumors at Johns Hopkins, and on laboratory experiments in mice, says Elana Fertig, Ph.D., professor and director of the Institute for Genome Sciences at the University of Maryland School of Medicine. Fertig co-led the project, beginning in her previous role at the Johns Hopkins Kimmel Cancer Center and continuing in her current role.

In one experiment designed to validate the new program, co-first author, Jeanette Johnson, Ph.D., a postdoctoral fellow at the Institute for Genome Sciences and recent graduate of the Immunology Ph.D. program at Johns Hopkins, ran the model to simulate how macrophages, a type of immune cell, invaded breast tumors by increasing expression of a genetic pathway called EGFR. Increasing this pathway typically promotes cancer growth. The simulation showed that tumors grew because cancer cells increased their ability to move.

With live breast cancer cells grown in the laboratory, the researchers observed the same type of tumor growth linked to an increase in cell movement.

“We still have a lot of work to do to add more cell behavior data to the program,” says Johnson, who is continuing this work as postdoctoral fellow with Fertig at the University of Maryland School of Medicine.

“We’re thinking of this project in terms of a virtual cell laboratory,” says Stein-O’Brien. Instead of doing all the experiments from the outset at the laboratory bench with living cells, the goal is to use these tools, which eventually could work as a “digital twin,” to prioritize hypotheses and therapeutic targets. “Then,” she says, “we can focus our benchwork on what seems most promising.”

In ongoing work, the team is using artificial intelligence to write simulation models using the new grammar, opening new possibilities for connecting models to new data and allowing medical research to improve digital twin models.