Skip to main content

Have you used our open science resources?


Simulations of life — in health and disease

Computational models allow us to ask: ‘What if?’” said Martin Meier-Schellersheim, Ph.D., a computational biologist at the National Institute of Allergy and Infectious Diseases, speaking last week at the Allen Institute to a large audience of modeling researchers.


5 min read

Peter Sorger, Ph.D., a Professor of Systems Pharmacology at Harvard Medical School, speaks at the recent Exploring Frontiers: Predicting Biology symposium at the Allen Institute.

Meier-Schellersheim was one of several researchers from around the world who participated in a symposium on modeling in biology, Exploring Frontiers: Predicting Biology. The two-day event brought together researchers who use computational simulations to better understand living systems, from the tiny ruffling edge of a migrating immune cell to how a large group of cancer patients respond to an experimental therapy.

The speakers at last week’s symposium, which was hosted by The Paul G. Allen Frontiers Group, a division of the Allen Institute, all use computational models of the cells, organs or populations they study to help them answer questions in a way that’s not possible with laboratory experiments alone.

“Models can help us get at what the truth really is,” said Markus Covert, Ph.D., director of the Allen Discovery Center at Stanford University and organizer of the event.

Covert described his group’s recent efforts to build an incredibly detailed computational simulation of the commonly studied bacteria E. coli, which they hope to debut later this year. Many models focus on one aspect or detail of a cell or living creature. The Allen Discovery Center team aims to wrap their arms around every known fact about this tiny bacterium, mining scientific publications for bacterial datasets of all types and recapitulating them all in an online software recreation of a complete bacterial cell that grows, divides, and responds to its environment.

But computer models aren’t the be-all and end-all solution to a research question.

“They’re not meant to be the way to predict everything or understand everything,” said Kathryn Richmond, Ph.D., Director of the Frontiers Group. “Models are really a tool to take complex data and elucidate parts of the issue. They need to be iterative, and they need to incorporate experimental data in those iterations.”

A model cell

The symposium was divided into three research sections that tied with the Allen Institute’s three scientific divisions: Cell Science, Brain Science and Immunology. In the cell biology modeling group, several other speakers touched on computational models of bacteria — especially in relation to infectious bacteria that impact human health.

Mary Dunlop, Ph.D., a bioengineer at Boston University, described her team’s research on how variation between individual bacterial cells can lead to antibiotic resistance. Dunlop’s computational models allowed the researchers to pinpoint how bacteria make tradeoffs between quick growth and protecting themselves against drugs.

Zaida Luthey-Schulten, Ph.D., a chemist at the University of Illinois at Urbana-Champaign, shared her computational modeling research to recreate the chemical reactions inside the “minimal cell,” a pared-down version of a bacterium with just 493 genes.

In some settings, computational modeling has to take into account both the microscopic and macroscopic scales, said Mark Brynildsen, Ph.D., a bioengineer at Princeton University. His research group is looking to combat antibiotic resistance from two angles: by boosting current antibiotics to better effectiveness and by looking for new therapies that render bacteria more vulnerable to our own immune systems.

“We’re looking at cell-to-cell combat, but we really want to predict what will happen at the population level,” Brynildsen said. “Who will win, the host or the bacteria?”

Capturing the brain’s complexity

Researchers speaking in the neuroscience modeling section also tackled the question of scales — from individual molecules to synapses to entire sections of the brain. It’s an ongoing challenge to understand the level of detail needed to answer a thorny question in neuroscience, the researchers said. But it’s not an impossible task.

Gaute Einevoll, Ph.D., a computational neuroscientist at the Norwegian University of Life Sciences, drew a parallel between modeling the brain and weather forecasts. Predicting the weather also relies on computational models of different scales to tackle a complicated problem.

“The key here is the use of mathematics,” Einevoll said. “That’s the only language that’s precise enough to link these scales.”

Adrienne Fairhall, Ph.D., a computational neuroscientist at the University of Washington, talked about how theory can guide both computational and experimental work to understand the brain. Fairhall is part of an advisory committee with the BRAIN Initiative, a National Institutes of Health-led neuroscience research endeavor with investigative groups all over the country, including at the Allen Institute.

In the BRAIN Initiative, the computational theorists are tasked with identifying fundamental principles of the brain. “Theory is really the heart of it,” Fairhall said. “Without theory, all the data we collect from these cool new tools is useless to us.”

Simulating disease

Several speakers at the event touched on models that could impact human health. Peter Sorger, Ph.D., a systems biologist at Harvard Medical School, described modeling and experimental work studying combination therapy for cancer patients. Sorger and his collaborators used mouse models of individual cancer patients’ and mined large datasets from clinical trials of combinations of drugs to come up with a surprising bottom line: Many so-called success stories of combination drugs can be explained by variation in the patient population.

That is, for the combination therapies Sorger and his team analyzed, it’s not that a single patient gets more benefit from taking two drugs at once, but rather that one patient in the trial benefited from one drug and another patient from the other drug.

Shai Shen-Orr, Ph.D., an immunologist at Technion-Israel Institute of Technology, also described ways that studying individual people rather than averaging datasets can lead to new insight for human health. By following changes in the immune system over time in individual people, Shen-Orr and his colleagues have devised ways to track a person’s “immune age.”

“For the level and scale of data the world is generating, we need to change how we model,” Shen-Orr said.

Videos of several of the talks from the event are available here.

Science Programs at Allen Institute