Recent advances in artificial intelligence, combined with vast experimental data on human biology, have converged to present a remarkable opportunity for scientific advancement.
Researchers from Stanford University, Genentech, and the Chan-Zuckerberg Initiative believe that science is now facing an “unprecedented opportunity” to harness AI to create the first-ever virtual human cell. This cell would have the capability to accurately represent and simulate how human biomolecules, cells, and ultimately, tissues and organs behave.
“Modeling human cells can be considered the holy grail of biology,” said Emma Lundberg, associate professor of bioengineering and pathology in the schools of Engineering and Medicine at Stanford. “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.”
A synthetic cell model could enhance our understanding of the intricate interactions among chemical, electrical, mechanical, and other factors that enable healthy human cells to function while also identifying the underlying causes of diseases that lead to cell impairment or death.
Even more fascinating is the prospect of an AI virtual cell, which would enable scientists to conduct experiments in silico—on computers instead of using live cells and organisms. This revolutionary approach could significantly enhance our comprehension of human biology while accelerating the search for new therapies, pharmaceuticals, and perhaps cures to disease.
Cancer researchers could simulate how specific mutations can transform healthy cells into malignant ones. Microbiologists might eventually forecast the impacts of viruses on infected cells and even their host organisms. Doctors could someday evaluate treatments using “digital twins” of their patients, hastening the arrival of a long-anticipated era of quicker, more cost-efficient, and safer personalized medicine.
However, according to the authors, for an AI virtual cell to be deemed successful, it must achieve three objectives. First, it should enable researchers to develop universal representations across different species and cell types.
Additionally, it must accurately forecast cellular functions, behaviors, and dynamics while understanding cellular mechanisms. Lastly, the AI virtual cell should facilitate computer-based experiments for testing hypotheses and directing data collection, enhancing its capabilities at a pace and cost much lower than what is currently possible.
According to the authors, we are entering a “trifecta” for science as AI has brought about tools that are predictive, generative, and query-able, yet the tremendous amount of raw biological data necessary to build the virtual cell is irrefutable.
In comparison, the authors highlight the Short Read Archive, a repository of DNA sequencing data from the National Institutes of Health, which now houses over 14 petabytes of data—this is a thousand times the size of the dataset used to train ChatGPT.
The journey to realizing the AI virtual cell will undoubtedly be challenging. It demands an unprecedented, collaborative effort on a global scale, involving open science across diverse fields, from genetics and proteomics to medical imaging, while fostering close partnerships among stakeholders in academia, industry, and non-profits.
Additionally, the authors emphasize that any progress toward the AI virtual cell should be pursued with the understanding that resulting models must be accessible to the entire scientific community without restrictions.
“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,” Lundberg asserted. “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.”
Journal reference:
- Charlotte Bunne et al. How to build the virtual cell with artificial intelligence: Priorities and opportunities. Cell, 2024; DOI: 10.1016/j.cell.2024.11.015