Researchers Create a Revolutionary Cell Atlas With AI

By using AI to analyze and learn from massive amounts of data, researchers have achieved remarkable breakthroughs in biology, from discovering new cell types to predicting genetic functions. This technology has the potential to revolutionize the way we explore and understand li

Share
Researchers Create a Revolutionary Cell Atlas With AI

In the late 1800s, French physician François-Gilbert Viault discovered that the human body can increase red blood cell production when needed, a phenomenon that led to the discovery of the hormone erythropoietin decades later. Research in this area took an exciting turn when biologists in Israel identified a rare kidney cell, known as the Norn cell, that produces the hormone during oxygen deprivation. This discovery took humans 134 years to make, but computers in California achieved the same feat in just six weeks last summer.

Stanford University researchers programmed computers to learn biology on their own by training them on raw data about millions of real cells and their chemical and genetic makeup. Without being given any predefined meanings for the data, the computers were able to build a model of all cells based on their similarities to each other in a vast, multidimensional space. Among the more than 1,000 cell types identified was the Norn cell.

New AI-Powered Programs Are Revolutionizing Biology

This software is one of several new AI-powered programs, known as foundation models, now exploring the fundamentals of biology. These models not only compile existing biological information but also make discoveries about gene function and cell development. With access to increasing amounts of lab data and computing power, researchers predict these models will make even deeper discoveries, potentially uncovering secrets about cancer and other diseases, as well as finding ways to turn one cell type into another.

From Cell Atlases to Mathematical Cell Models

Researchers have started using industrial-scale experiments to identify genetic components in individual cells, leading to the creation of "cell atlases" containing billions of data points. Christina Theodoris at Boston Children's Hospital was inspired by an AI model created for language translation and wondered whether a similar model could be used to interpret data in cell atlases. After pushing through skepticism, she managed to build a model that provided a deep understanding of how our genes behave in different cells.

The Virtual Cell: A Revolution in Biology

With enough data and computing power, researchers may eventually be able to create a complete mathematical representation of a cell, a development that would be revolutionary for biology. This "virtual cell" would allow researchers to predict how a real cell would react in any conceivable situation, and potentially create new cell types that don't exist in nature today.

Challenges and Risks of Future Research

Despite the major advances and potential, this technology also raises new risks, including the possibility of creating biological weapons and privacy breaches through personalized models. Skeptics also question how far these models can go in terms of fundamental new discoveries, since they depend on the data they're fed.

In summary, we're facing a new era in biological research, where AI's ability to rapidly identify and classify cell types and genetic functions could lead to groundbreaking discoveries. This development challenges traditional methods and redefines our understanding of life at the cellular level.