When Xiaoyu Shi’s grandmother was diagnosed with Alzheimer’s disease, a question she couldn’t shake stayed with her. As a postdoc at UCSF, she began pressing colleagues: why, after decades of research, was there still no cure? When she accepted a faculty position at UC Irvine, home to one of 37 federally designated Alzheimer’s Disease Research Centers, she kept asking. The answer was always the same: the science wasn’t there yet.

For decades, research on Alzheimer’s treatment had centered on beta-amyloid plaques, the protein clusters that accumulate in the brain during late-stage disease. That focus produced important advances, but it also revealed a gap. By the time plaques appear, the disease has often been underway for years, driven by earlier molecular changes that no one could reliably detect. The precursors existed. The technology to find them did not.

“I realized I had a tool that could help,” Shi says. “Detecting low-abundance proteins is exactly what my lab’s GEMclear proteomics was invented for.”

That realization is what led her to start Cytoseen, the biotech startup Shi, now an associate professor of Developmental and Cell Biology at UC Irvine, with joint appointments in Biomedical Engineering and Chemistry, co-founded in 2025 with her CEO, Yang Liu.

The 90 Percent We Can’t See
To understand what Cytoseen is building, it helps to understand a foundational gap in modern medicine. The human genome encodes roughly 20,000 protein-coding genes, which give rise to a far larger set of distinct proteins. Proteins are the workhorses of biology. They carry signals, trigger immune responses, and tell cells when to divide and when to stop. They are also what most of the drugs are designed to target.

And yet, current pharmaceutical research focuses on a small fraction of them. By most estimates, only 10 to 15 percent are well-studied, and FDA-approved drugs target closer to 3 percent. The remaining 90 percent are largely understudied, not because scientists have ignored them, but because the tools to study them have been inadequate. Many of the most consequential proteins, the ones governing disease at its earliest, most treatable stages, are present in such small quantities that standard detection methods miss them.

To understand what these low-abundance proteins mean to health, Shi suggests thinking of traffic. Most proteins are like the cars on the road, abundant, constantly moving, easy to count. But what actually keeps traffic flowing is a much smaller, less visible system: the signals at each intersection. When the lights work, you barely notice them. When one fails, the whole network can grind to a halt. In the body, signaling proteins play that role. They are rare, decisive, and disease-specific. And until now, they have been exceptionally difficult to detect reliably.

Shi’s GEMclear proteomics technology works in two steps. First, the system chemically tags proteins at key signaling hubs, where the cell processes information and makes decisions. Then it removes the abundant background proteins that would otherwise drown out the signal. What remains is sent to a mass spectrometer, an instrument that identifies proteins the way a fingerprint identifies a person.

This is a meaningful distinction from what came before. Most current detection systems rely on antibodies, which can only detect antibody-targeted proteins, but not for the unbiased exploration of new drug targets. Cytoseen’s GEMclear platform can detect proteins researchers didn’t even know to look for, plus it’s 10 to 20 times more sensitive than existing methods.

The Right Partner
Shi and Liu met about a decade ago through a mutual friend. For years, their relationship was purely personal. They were two scientists navigating academic careers, both on the tenure track, both focused on publishing rather than commercializing their research. That changed around 2020, when Liu walked away from her tenure-track position. Shi’s first reaction was disbelief. ‘I thought Liu must have a great reason that I don’t know,’ she says. That great reason turned out to be entrepreneurship.

Liu had been trained as a pharmaceutical scientist, earning her Ph.D. at the University of North Carolina, Chapel Hill. She spent five years as a founding faculty member at Chapman University in Orange, California. But she had come to believe that academia, for all its rewards, was not the right setting for the kind of work she wanted to do. “Business rewards execution,” she says. “That’s where my strengths are.” She left her tenure-track position and spent several years working on a cloud-based services company, where she learned the mechanics of building something from the ground up.

Liu describes herself as someone who takes calculated risks, especially when the downside is capped and the upside could be life-changing. She and Shi sat down and mapped out what each of them wanted, what they valued, and where they might conflict. Even with that preparation, starting the company still took a leap of faith.

When Shi learned that Liu had left academia for entrepreneurship, she saw her friend in a new light. The two women were deliberate in deciding whether to build something together. What they recognized was not just shared ambition, but years of trust, a clear understanding of each other’s values, and complementary strengths shaped by different career paths. It was a foundation they believed would translate beyond friendship into a working partnership.

“People say, if you want to start a company, don’t go into business with your friends,” Shi says. “I disagree. It’s more important to choose the right person.”

Their initial focus is cancer and neurodegenerative diseases like Alzheimer’s. Within cancer, they are zeroing in on the environment surrounding a tumor, which largely determines how a cancer grows and whether treatment works.

The choices were not arbitrary. “Cancer and neurodegenerative disease still put the most burden on our aging people and on our entire healthcare system,” says Liu. “There are huge unmet needs.”