In a low building off Oyster Point Boulevard, a small team is doing something most of the drug industry still treats as heresy: it skips the mouse. Verge Genomics starts with human samples, hands them to a machine, and waits for biology to confess.
A 26-person bet that human data beats the mouse
Verge Genomics is a South San Francisco biotech with an unfashionable conviction: the fastest way to a working drug is to start with the species you intend to cure. Most discovery pipelines begin in mice or cultured cells and hope the biology travels. Verge built one of the largest proprietary multi-omics datasets in the field straight from human brain and tissue, then pointed machine learning at it to find what actually drives disease.
Today the company is leaner and more focused than it was at its peak. After a hard year, it has narrowed to the thing it always did best - the platform. The drugs were the headline. The data was the company.
A biotech the size of a dinner party, sitting on a dataset the size of a small country's medical archive.
“We use human data and AI to develop better drugs, faster.”
// Verge Genomics, mission statementNeurodegeneration is where drug programs go to die
ALS, frontotemporal dementia, Parkinson's - these diseases have humbled the pharmaceutical industry for decades. Candidate after candidate looks brilliant in a mouse and does nothing in a person. The standard explanation is bad luck. Verge's founders suspected it was a bad starting point.
A mouse engineered to mimic ALS is not a person with ALS. The translation step - from animal model to human patient - is where billions of dollars quietly evaporate. The industry treats that gap as a cost of doing business. Verge treated it as the bug to fix.
“Instead of animal or cell models, Verge maps disease directly from human tissue.”
// The all-in-human premiseShe left her MD/PhD three months before graduation
Alice Zhang was deep in the UCLA-Caltech MD/PhD program, studying gene networks in neuro-regeneration, when she walked away - three months short of the degree - to start Verge in 2015. Her co-founder, Jason Chen, came out of the same world of systems biology and computational research. Their thesis was almost rude in its simplicity: the field had the human data and the machine learning; it just refused to put them in the same room.
So Verge assembled a team that mixed two tribes who rarely share a lunch table - machine-learning engineers and seasoned neuroscience drug developers - and built CONVERGE, a platform whose name nods to the collision of human genomics, machine learning, biological engineering, and translational medicine.
Neuroscientist-turned-founder who left UCLA's MD/PhD program to bet that AI plus human biology could change drug discovery.
Systems-biology researcher with a background spanning Duke and UCLA, working on the genomics of neurodegenerative disease.
Two researchers, one shared suspicion that the entire industry was looking in the wrong species.
“A young founder's superpower is being willing to question what everyone else accepts.”
// On Alice Zhang's path, BioPharma DiveThe Verge timeline
A decade in seven beats. Each dot is a bet that mostly paid off.
- 2015Alice Zhang and Jason Chen found Verge Genomics.
- 2018$32M Series A to scale AI-driven drug discovery.
- 2021 - JULEli Lilly collaboration on ALS, up to ~$719M including milestones.
- 2021 - DEC$98M Series B led by BlackRock; Lilly and Merck invest.
- 2022VRG50635 enters the clinic; Verge named to the Fierce 15.
- 2025ALS candidate fails its proof-of-concept trial.
- 2025Company refocuses on the platform, repositioning as a data/partner business.
CONVERGE: a map of disease drawn from real patients
The heart of Verge is its dataset. Rather than buying biology off the shelf, the company built a proprietary multi-modal molecular and clinical archive derived directly from patient tissue - one of the largest of its kind. CONVERGE mines that archive to identify which targets actually cause disease and which drug candidates are most likely to work in humans.
The proof that the approach was not just a pitch deck: VRG50635, a small-molecule PIKfyve inhibitor for ALS, went from research to the clinic in roughly four years. In an industry that measures discovery in decades, that pace turned heads.
The drug had the company's initials in its name. Biotech is rarely that on-the-nose.
“From research to clinic in four years - in a business that usually counts in decades.”
// On VRG50635's paceBig money read the same data and leaned in
Conviction is cheap; a $98M oversubscribed round is not. In December 2021, BlackRock led Verge's Series B, joined by Eli Lilly, Merck's Global Health Innovation Fund, Section 32, and Vulcan Capital. Months earlier, Lilly had signed a three-year ALS collaboration worth $25M upfront and up to roughly $694M in milestones, with the option to advance up to four candidates.
Funding, stacked
Verge's capital story, in millions of dollars. Bars animate on scroll.
Figures from public announcements. Lilly's deal also carried up to ~$694M in potential milestones, not shown here.
Then biology had the last word. In its proof-of-concept trial, VRG50635 failed to help patients - neurofilament light chain, a marker of nerve damage, went up rather than down. Verge dropped its only clinical candidate, laid off roughly 90% of staff, and turned back toward the asset that had carried it the whole way.
The cruelest part of drug development: the dataset can be right and the molecule still wrong.
“The drugs were the headline. The platform was the company.”
// The logic behind the 2025 refocusAutomate discovery, or at least stop guessing
Verge's founding vision was audacious: become the first pharmaceutical company to automate its drug discovery process end to end. The 2025 pivot doesn't retire that ambition so much as relocate it. Instead of carrying every molecule to market itself, Verge now aims to supply other drug developers with human-grounded target data - selling the map rather than walking every road.
It is a more modest business and, arguably, a more honest one. The platform's whole promise was that human data predicts human outcomes better than a mouse ever could. Letting partners test that claim across many programs is, if anything, a bigger experiment than running one trial alone.
“The vision: the first pharma company to automate drug discovery itself.”
// Verge's founding ambitionThe next ALS drug may start in a human, not a mouse
Whether or not Verge itself delivers the cure, the premise it has spent a decade defending is now spreading through the industry: begin with human biology, let machines find the pattern, and treat the animal model as a checkpoint rather than a starting line. That shift outlives any single failed trial.
For patients facing diseases that have defeated every prior approach, the stakes are not abstract. A faster, more human path to candidates is the difference between a drug arriving in time and arriving too late.
Back in that low building off Oyster Point, the team is smaller now, the lab quieter. But the conviction that started it is intact: hand human biology to a machine, and wait for the answer. Verge bet the mouse was the problem. The rest of the industry is starting to agree.