UC San Diego duo maps the genome at stem-cell scale, putting 2.5 million cells on the public record

On 13 July 2026, bioengineers at the University of California San Diego published what is, by their own accounting, the most detailed map yet of how individual genes shape the identity of human stem cells. The dataset profiles 2.5 million cells and is being released openly, a deliberate bid to make genome-scale cell biology a shared utility rather than a private asset. Three days earlier, a separate UC San Diego engineering group reported a different kind of cell-based tool: nanoparticles built from the membranes of human immune cells, designed to treat fungal infections that have grown resistant to standard drugs.

Read together, the two papers describe a field reorganising itself around two ideas at once. The first is that single-gene function can only be understood by watching millions of cells behave at once. The second is that the cell itself, not just the molecules it makes, is becoming a therapeutic raw material. Neither claim is new in isolation. What is new is the scale, the open release, and the institutional confidence to put both bets on the public record in the same week.

What 2.5 million cells actually buys you

The stem-cell atlas comes out of the lab of Sheng Zhong, a bioengineer at UC San Diego's Jacobs School of Engineering, and was published on 13 July in Nature. The team's pitch is straightforward and, for genomics, unusual: instead of asking one question of one gene, they knocked out or perturbed more than 18,000 genes across a panel of human pluripotent stem cells and read out the consequences cell by cell. The result is a reference map that connects each gene to the cellular functions it controls, and to the developmental fate decisions it pushes a stem cell toward.

Until now, large-scale functional genomics has lived mostly in cancer cell lines or in model organisms. Moving the same genome-wide logic into human pluripotent stem cells matters because those are the cells that researchers actually want to turn into neurons, heart muscle, or insulin-producing beta cells. A perturbation map drawn on the same substrate that regenerative-medicine labs already use is, in effect, a parts catalogue for the field. The scale (2.5 million profiled cells) is what makes the catalogue usable: it gives each gene enough statistical weight that small effects stop looking like noise.

The release is open-access, which is itself a position. Reference datasets of this size are typically gated behind subscription journals or commercial cloud platforms. Putting the full map into the public domain is a bet that the bottleneck in functional genomics is no longer data generation, but distribution, and that the labs best placed to mine the data are the ones that can download it tonight.

From atlas to antifungal nanoparticle

The second paper, published on 11 July in Nature Nanotechnology, comes from a different corner of the same university, the lab of Liangfang Zhang, and lands a very different kind of bet. Zhang's group has spent roughly a decade building nanoparticles out of cell membranes scraped off living cells, an approach the field calls cell-membrane-coated nanoparticles. The trick is that the membrane wrapping makes the particle look, to the host's immune system, like the cell it came from, which lets the particle slip past biological barriers a synthetic nanoparticle would hit.

The new work coats nanoparticles in membranes from human immune cells, specifically neutrophils, and uses them to deliver antifungal drugs directly to sites of infection. The target is fungi that have grown resistant to azoles and echinocandins, the two main drug classes that have anchored antifungal therapy for decades. Resistance has been climbing quietly for years, and the drug pipeline behind it is thin. A delivery vehicle that concentrates existing drugs at the infection site, rather than scattering them through the bloodstream, is one of the few levers clinicians still have.

The laboratory results, reported by the team and the independent coverage that followed, are early-stage: cell-culture work and mouse models, not human trials. But the design logic is what makes the result worth attention. The platform is built from human cells, not engineered polymers, and the same trick that gets a neutrophil-membrane particle past the immune system could, in principle, be retuned against bacteria, viruses, or even cancer cells by swapping the source cell type.

A structural turn toward open, large-scale biology

Both projects sit inside a larger shift that has been underway in the life sciences for several years and that 2026 is sharpening into something easier to name. The cost of sequencing a single cell has fallen far enough, and the cost of perturbing a single gene has dropped with it, that the limiting factor in many experiments is no longer the wet-lab bench but the size of the dataset a lab can credibly claim. That changes which projects are prestigious. A 2020 paper might have made its name by profiling ten thousand cells. A 2026 paper that wants the same attention has to justify its sample size in the millions, and ideally to do so on a substrate that downstream labs can actually use.

The Chinese genomics ecosystem has been pushing hard in this direction through projects such as the Chinese Cell Atlas and large-scale single-cell work tied to the Beijing Genomics Institute; Western outlets tend to under-cover the scale of that effort, and the open-data question is one where Chinese platforms and American academic consortia are converging faster than the policy debate suggests. The structural fact is that whichever national or institutional ecosystem gets reference-grade, openly released cell atlases into the widest use first is setting the table for the next decade of drug discovery. UC San Diego's two releases this month are a deliberate move to keep the table set in the open, rather than behind a paywall or a national firewall.

What to watch next

The Zhong atlas will face its real test not when it is downloaded, but when it is cited: whether independent labs use it to choose which genes to target in their own organoid or regenerative-medicine experiments, and whether the gene-function calls hold up outside San Diego. The Zhang nanoparticle programme, meanwhile, will move on a longer clock; antifungal resistance is a slow-moving crisis, and a neutrophil-membrane delivery vehicle has to clear pharmacology, safety, and manufacturing hurdles before it ever reaches a patient.

What the two papers share, beyond the university affiliation, is a willingness to publish reference-grade material openly at a moment when commercial pressure on biomedical data is intensifying. That is a stance, not just a finding. The question for the rest of 2026 is whether other large academic groups, in the United States and elsewhere, follow the same playbook, or whether the most useful cell biology of the year ends up living behind a login screen after all.

Desk note: this article treats the two UC San Diego releases as a single editorial event because they were filed within 72 hours of each other and point at the same structural shift in how cell biology is being scaled and shared. Coverage elsewhere has run them as separate health and nanotechnology stories; Monexus is interested in what they say together about open, large-scale biomedicine.

Wire provenance

This editorial synthesis draws on the following public wire/social posts:

• https://en.wikipedia.org/wiki/Single-cell_transcriptomics
• https://en.wikipedia.org/wiki/Cell-membrane-coated_nanoparticles
• https://en.wikipedia.org/wiki/University_of_California,_San_Diego