Jason Zhang designs proteins that have never existed in nature. The UCLA bioengineer, who joined the faculty in 2025 after completing a postdoc in Nobel laureate David Baker's lab at the University of Washington, uses generative AI to invent proteins for drug development, disease research, and a future "virtual cell" that could predict how therapies affect human tissue.
Zhang is a member of the California NanoSystems Institute at UCLA and the UCLA Health Jonsson Comprehensive Cancer Center. His methods represent a shift from traditional protein engineering, where researchers laboriously tweak naturally occurring proteins, to a generative approach where AI models create amino acid sequences from scratch.
"I could see the promise of using a computer to dream up proteins never seen in biology," Zhang said. "It's an exciting field that is still very much in its infancy, but I am confident it is the future."
How AI protein design works
Zhang's lab trains AI models on experimental data, then specifies a target for the protein to bind to. The AI generates candidate proteins and outputs their amino acid sequences. The team reverse-translates the sequences into DNA and expresses the proteins in a wet lab.
The efficiency gain is dramatic. A single experiment can test 20,000 designed proteins for binding activity. Promising candidates are then tested individually, followed by experiments in cells and disease models.
The underlying approach is connected to broader courses for professionals working in scientific fields. The methods Zhang uses have applications in biotech research, pharmaceutical development, and academic biology, where AI is becoming a standard tool.
Targeting the undruggable
Much of Zhang's lab focuses on disordered proteins, which drive neurodegenerative diseases, diabetes, and some cancers. These proteins are notoriously hard to target with small-molecule drugs because they lack a stable 3D structure - no predictable "pocket" exists for a drug to fit into.
"So we use generative AI to create totally new folds meant to drug the undruggable," Zhang said. "The binder we create has the pocket, essentially, that the disordered protein fits into."
A significant portion of his translational work happens through collaborations. His team builds diagnostic molecules and tools for drug development. One project involves using CAR-T cells, an immunotherapy that has succeeded against blood cancers, to target a rare liver cancer.
A longer-term vision: the virtual cell
Zhang's broader research goal extends beyond individual molecules. "The virtual cell is a big vision I'm interested in," he said. "If these new tools, such as biosensors, can generate a lot of data to profile cells, we might be able to create a digital twin. Maybe one day, an AI model can predict how various drugs affect, say, immune cells or cancer cells."
He is also working toward personalized medicine. As disease classification becomes more granular, he said, a single disease category can break into many distinct conditions. If his methods prove effective, therapy could be individualized per patient.
Zhang's work aligns emerging AI and biotech developments with fields like laboratory research and biomedical design.
Why the UCLA setting matters
Zhang chose UCLA in part for its public mission. "I've been at public schools my entire academic career, from elementary school back to Virginia through my postdoc at University of Washington," he said. "And now I'm at UCLA, where we educate people of all backgrounds, primarily focusing on the state of California, as well as providing an engine of innovation."
He also cited the institutional environment, noting his lab sits "in between the medical school and the engineering school" with access to shared facilities and staff support.
His research has received an NIH Pathway to Independence Award, which funded his transition from postdoc to principal investigator. "It's a comfort to know I have resources to be paid my team and get the lab running," he said.
Why this matters for science and research professionals
Zhang's research shows how experimental science is changing: a PhD project that took years of slow, difficult protein engineering can now be attempted with AI generating thousands of candidates in days. For researchers deciding whether to invest in these methods, the lesson is that AI protein design tracks already outpace what biology can produce through natural evolution.
Your membership also unlocks:
