Company Profile
Keythera Pharmaceuticals Keythera (Suzhou) Biopharmaceuticals Co., Ltd.
A small lab in Suzhou aiming a mass spectrometer at two of medicine's oldest opponents - cancer and the immune system that turns on itself.
A small lab in Suzhou aiming a mass spectrometer at two of medicine's oldest opponents - cancer and the immune system that turns on itself.
In a unit at BioBay in Suzhou, a workstation is sorting through chemistry at a pace no human chemist could match. Two hundred and fifty thousand compounds a day. Each one a tiny question put to a protein: do you bind, or do you not? Most do not. That is the whole point. Keythera is in the business of finding the few that do, and the few among those that might one day become medicine.
This is Keythera (Suzhou) Biopharmaceuticals - a clinical-stage biotech with a tidy team, six drug pipelines, and one unfashionable conviction: that the fastest way to find a first-in-class small molecule is to measure binding directly, not to predict it and hope. The company sits at the intersection everyone in 2026 claims to occupy - AI, drug discovery, oncology - but it got there by way of a physics instrument that has been around for a century.
Here is the uncomfortable arithmetic of new medicine. A promising target - a protein implicated in a tumor's growth, say - can in principle be hit by an astronomical number of small molecules. Finding the right one has traditionally meant testing candidates one assay at a time, slowly, expensively, and with a failure rate that would end most other industries. The needle is real. The haystack is the size of a country.
The fashionable answer this decade is to let an algorithm guess where the needle is. Useful, sometimes. But a prediction is not a measurement, and a model that has never touched the protein is, in the end, an opinion. Keythera's founders had spent careers watching confident opinions fail in the clinic. They wanted data that argued back.
Dr. Yongqi Deng did not need to start a company. He had the kind of resume that usually ends in a corner office: bachelor's and master's degrees from Peking University, a PhD from Michigan State, a postdoctoral fellowship at MIT, and more than two decades of drug research at Schering-Plough and Merck (MSD). He had seen, from the inside, both how Big Pharma finds drugs and how often it does not.
In 2018 he returned to China and founded Keythera in Shenzhen. In May 2020 he moved the operation to Suzhou's BioBay - the country's most concentrated biotech cluster, the kind of place where your neighbors are also trying to cure something. The bet was specific: build a screening engine fast and accurate enough that a lean team could chase first-in-class targets that larger, slower companies tend to avoid.
That measured binding affinity, at industrial throughput, beats prediction alone - and lets a small team punch above its headcount.
First-in-class is the hardest game in the field. There is no competitor's drug to copy and improve. You are first, or you are nothing.
Above: the founders' two-by-two. Biotech rarely fits on an index card, but the strategy does.
The engine is called ADMS - Affinity Detection by Mass Spectroscopy. The name is a mouthful; the idea is not. Mass spectrometry measures the mass of things with extraordinary precision. Point it at a protein mixed with a library of compounds, and you can detect which compounds are actually stuck to the protein - which ones bind. No fluorescent tag, no proxy signal, no guesswork. The instrument simply reports what is there.
Around that core, Keythera layers the rest of the modern toolkit: bioinformatics, chemical informatics, structural chemistry, computer-aided drug design, combinatorial chemistry, and, yes, artificial intelligence. The difference is the order of operations. The AI here is fed a steady diet of real experimental binding data, so its suggestions are checked against the bench rather than substituting for it. Keythera describes this as overcoming the limits of AI-only approaches - the model proposes, the spectrometer disposes.
The throughput is the headline: up to 250,000 compounds screened per day on each ADMS workstation. The promise underneath it is subtler - faster hits, fewer dead ends, and a lower chance of pouring years into a molecule that was never going to bind in the first place.
Biotech is full of beautiful slide decks and empty pipelines. So the fair question for any platform company is: did it produce anything? Keythera's answer is a portfolio of six pipelines aimed at cancer and autoimmune disease, led by KF-0210, an EP4 receptor antagonist for solid tumors that has moved into early clinical development. EP4 is a target with real biological rationale in oncology - the sort of program a platform is supposed to generate.
Investors put money behind the thesis. The Series A of nearly RMB 100M was led by Oriza Seed, with Huaige Capital, Ming Bioventures, and Shangyi Guanxi Capital following. The capital was earmarked for the first anti-tumor IND application and for scaling the screening-and-optimization platform - in other words, for turning throughput into pipeline.
Bars are directional, not audited - they illustrate the order-of-magnitude gap Keythera's platform is built to exploit. The 250K/day figure is the company's own stated capacity per ADMS workstation.
Plenty of companies chase "me-too" drugs - safer bets that improve on a molecule someone else already proved. Keythera states a harder goal: first-in-class small molecules for cancers and autoimmune disorders, aimed at unmet medical need. Its stated vision is to become a frontrunner in groundbreaking small molecule discovery within China while building a company with a global outlook. Its values, the company says, are innovation and integrity - the second of which is the one most biotechs forget to mention.
The culture is built to match. Nearly nine in ten staff work in R&D. Everyone holds at least a bachelor's degree; almost a fifth hold doctorates. It is, by design, a company of people who would rather run the experiment than argue about it.
The bet is not yet won. Drug discovery is a long game, and a Phase Ib program is a beginning, not a finish line. Plenty can still go wrong between a binding event in a spectrometer and a benefit in a patient. Keythera knows this better than most; its founder spent twenty years watching the gap claim good ideas.
But the logic is sound. If you can measure binding at a quarter-million compounds a day, and if you keep your AI honest by feeding it real data, then a four-to-thirty-person team can credibly run six programs against targets that big companies leave alone. That is the whole proposition - not a slogan, a workflow.
So return to that lab at BioBay. The workstation is still sorting chemistry, still asking its one tireless question - bind, or do not bind. Three years ago it was a thesis. Now it is feeding a pipeline, with a molecule already in the clinic and five more behind it. The machine has not changed. What it is finding has.
Note: figures and stages reflect publicly available sources and the company's own materials; some details (team size, exact clinical stage) vary across databases and should be treated as approximate.