NYC Gazette

Columbia Vagelos College of Physicians and Surgeons is making strides in systems biology with the aim of creating a virtual cell modeled on a computer. This innovative approach could replace traditional lab experiments by providing insights into biological responses to genetic mutations or experimental drugs.

Mohammed AlQuraishi, an assistant professor at Columbia's Vagelos College, believes that such an AI-powered simulation might be less than 15 years away. His current research focuses on using machine learning to predict protein folding, a longstanding challenge in biology. "If we can predict the structure of molecules, then we can next predict how molecular machines assemble," says AlQuraishi. "This would completely change how we study disease and design drugs."

AlQuraishi initially pursued computer science but shifted his focus to biology after discovering parallels between computer programming and cellular processes. He later studied genetics at Stanford and developed models for protein-DNA interactions. During his fellowship at Harvard Medical School, he turned to deep learning to address protein folding prediction—a problem that traditional methods struggled to solve.

While Google's AlphaFold made significant advances in this area, AlQuraishi saw opportunities for further development. He led the creation of OpenFold, an open-access AI tool designed to predict protein structures without AlphaFold's limitations. "AlphaFold did well for individual proteins but it worked less well for protein complexes," said AlQuraishi.

OpenFold has been adopted globally by researchers who use it in diverse ways, including integrating different experimental results to enhance predictions. As more data is incorporated into OpenFold and its successors, AlQuraishi envisions achieving increasingly complex predictions leading toward simulating an entire cell.

"I think we can’t ever be certain that we understand all the essential features of a living cell unless we can model it," he says. Originally predicting such technology by 2050, he now estimates its arrival by 2040 due to rapid advancements in the field.

"This field is moving at the speed of light," he remarked, urging people to consider recent progress and future possibilities as developments continue at a remarkable pace.