AI Creates the First Virtual Human Cell

A team of scientists from Stanford University, Genentech, and the Chan-Zuckerberg Initiative claims that science has an “unprecedented opportunity” to use artificial intelligence (AI) to create the first virtual human cell in history, pointing out that recent advancements in AI and the availability of extensive experimental data about human biology have reached a critical mass. The exact behavior of human macromolecules, cells, and eventually tissues and organs might be accurately represented and replicated by such a cell.

Modeling human cells can be considered the holy grail of biology. AI offers the ability to learn directly from data and to move beyond assumptions and hunches to discover the emergent properties of complex biological systems.”

Emma Lundberg, Study Senior Author and Associate Professor, Stanford University

Lundberg's senior authors include two Stanford colleagues, Stephen Quake, a bioengineering professor and head of science at the Chan-Zuckerberg Initiative, and Jure Leskovec, a computer science professor in the School of Engineering, as well as Theofanis Karaletsos, head of artificial intelligence for science at the Chan Zuckerberg Initiative, and Aviv Regev, executive vice president of research at Genentech.

Remarkable Promise

Such a synthetic cell model would provide a better understanding of the complex interplay of chemical, electrical, mechanical, and other forces and processes that keep healthy human cells functioning, as well as show the underlying causes of disease that cause cell dysfunction or death.

Perhaps more intriguingly, an AI virtual cell would allow scientists to experiment in silico rather than in vivo, that is, on a computer rather than on living cells or creatures. This skill would broaden human understanding of biology and accelerate the search for new therapies, medications, and, perhaps, disease cures.

Cancer researchers might create models that show how particular mutations cause healthy cells to become malignant.

Microbiologists might one day be able to forecast how viruses will affect infected cells and possibly host organisms. Physicians might one day test treatments on “digital twins” of their patients, ushering in a long-awaited era of faster, cheaper, and safer tailored medicine.

The authors believe an AI virtual cell must achieve three goals to be regarded as a success. First, it would need to give researchers the ability to generate universal representations across species and cells. It would also need to be able to reliably forecast cellular function, behavior, and dynamics and understand cellular mechanisms.

Finally, an AI virtual cell would enable computer experiments to test ideas and drive data collecting to increase the virtual cell’s capabilities at a far faster and lower cost than today.

Global Collaboration

In what the authors call a “trifecta” for research, AI has ushered in an era of predictive, generative, and query-able tools, but the huge amount of raw biological data required to generate the virtual cell cannot be denied.

In comparison, the authors point to the Short Read Archive, a repository of DNA sequencing data produced by the National Institutes of Health that already has more than 14 petabytes of data, which is a thousand times greater than the dataset used to train ChatGPT.

Creating the AI virtual cell will not be easy. It will require unprecedented worldwide open-science collaboration in domains ranging from proteomics and genetics to medical imaging, as well as close cooperation across global stakeholders in academia, industry, and non-profits.

At the same time, the authors emphasize that any work on the AI virtual cell should be done with the understanding that the resulting models will be freely available to the entire scientific community.

Lundberg concluded, “This is a mammoth project, comparable to the genome project, requiring collaboration across disciplines, industries, and nations, and we understand that fully functional models might not be available for a decade or more. But, with today’s rapidly expanding AI capabilities and our massive and growing datasets, the time is ripe for science to unite and begin the work of revolutionizing the way we understand and model biology.”