Nianzhen Li, Arc’s Senior Director of Multi-omics Technology Development, on developing the experimental platform for a virtual cell

Nianzhen Li’s team is trying to generate the largest single-cell perturbation dataset ever assembled. Those data will feed into Arc Institute’s first virtual cell model, called State, which predicts how cells’ gene expression profiles shift after a genetic or chemical perturbation. An accurate, predictive virtual cell model could help scientists run computational assays to figure out which drugs are likely to shift cells from “diseased” to “healthy” states, before ever running a wet-lab experiment. Building such a model will require collecting single-cell data on hundreds of millions of cells.

It’s an ambitious goal, but Li, who directs Arc’s Multi-Omics Technology Center, is unusually well-equipped to pull it off. Over the past two decades, she’s moved between academic labs and early-stage startups, designing microfluidic chips, building commercial instruments, and scaling up single-cell technologies long before they were mainstream.

Li started out in academia as a neuroscientist, but has always had a penchant for tool development. During her postdoc at the University of Washington, she joined a bioengineering lab that was among the first to apply microfluidics to neuroscience. “We were developing microfluidic devices to study axon guidance and synaptogenesis,” she says—the process by which neurons find their targets and wire together. She learned how to build custom chips in the lab, including one device that could physically stretch strands of DNA.

From there, she moved into industry and stayed for more than a decade. At Fluxion Biosciences, an early-stage company with just five employees when she joined, Li worked on automating patch-clamp assays—electrical recordings from single cells that are notoriously tricky to do by hand. “You had to wear many hats,” she says. “Hardware, software, working with biologists…I learned how to make tools that were actually useful.”

Later, at Fluidigm (Now Standard BioTools), she helped turn single-cell analysis into a commercial product. She led a small team developing an automated system for stem cell reprogramming, using nanoliter-scale chips to dose cells with combinations of Yamanaka factors. After building a series of prototypes, Li’s team actually launched a commercial product. Eventually, she took over Fluidigm’s broader single-cell platform efforts, including a system that could capture and process 800 cells at a time. “Now at Arc, I can do a million in a run,” she says, “but back then, 800 was a lot.”

She has always been interested in multiomics analysis of single cells. After Fluidigm, she led the droplet single cell DNA-sequencing chemistry development at Mission Bio and then built the biology team and helped to develop an AI enabled label free imaging single cell morphology-based sorting system at Deepcell. “I enjoy developing tools and technologies that are enabling the scientific community; however when you’re working on one or two products, every decision eventually comes down to commercial priorities,” she says. Li wanted to work on longer-term, riskier ideas that had real potential to move biology forward. So when Arc Institute began recruiting for its new Technology Centers in 2022, Li signed on as the first employee. The Arc model was compelling to Li, in part, because it pairs academic research laboratories with dedicated technology teams focused on longer-term, high-risk projects. Arc offered Li the intellectual freedom of academia without the pressure to secure funding or publish papers.

Today, her team includes a dozen scientists and engineers, split between two major responsibilities: platform and development. On the platform side, the Multi-Omics Technology Center runs the institute’s core sequencing machines, including high-throughput library prep for RNA, protein, cytokine, and live-cell imaging assays. On the development side, they build new workflows for Arc’s two major initiatives: the Virtual Cell Initiative (VCI) and Alzheimer’s Disease Initiative.

That means running large-scale perturbation screens while developing new single-cell assays. “We’re not building our own microfluidic devices,” she says. “Instead, we work with commercial platforms or collaborators and make them better; more sensitive and more scalable.”

Her group is also layering in additional readouts that will feed into future, multi-modal versions of Arc’s virtual cell model. “Ultimately, we want to combine CRISPR perturbations with in situ readouts,” Li says. “That means connecting morphology and protein levels and transcriptomics all into one workflow in tissue.”

In a sense, Li’s team is building the experimental backbone for Arc’s most ambitious ideas. They may not be making new chips or devices from scratch, but they are stitching together tools—wet lab, automation, and software—into workflows that actually work at scale.

Li has also helped to shape Arc’s virtual cells data strategy from the ground up. “Two years ago, it was just a concept,” she says. “We didn’t have the tools, we didn’t have the throughput—it felt like a dream.” But her team quickly built the experimental platforms to make it real and, by the end of this year, they expect to generate insights from more than 100M single cells.

“It’s hard,” she says, “but not impossible. We’ve made a lot of progress. If we stay aligned and focused, we can do it.”