
Chris Varma unveils MPM's latest startup — eyeing 'undruggable' cancer targets and powered by machine learning, $67M
Two years after MPM Capital enlisted Chris Varma on its busy oncology team, the former entrepreneur-in-residence is unveiling his first venture project out of his new stomping grounds in the Bay Area: Frontier Medicines.
For Varma, who’s also co-founded Blueprint Medicines and built companies at Third Rock and Flagship, this marks another opportunity to apply some cutting-edge science to “several of the most important and difficult targets in cancer” — targets that others have tried to tackle with more classical methods and failed. The launch round comes in at $67 million, which should go some way in scaffolding a preclinical pipeline and push one or more assets into the clinic three years from now, he tells me.

Frontier’s stated mission — drugging previously “undruggable” proteins — comes right out of MPM’s playbook for a $408 million fund that managing director Luke Evnin said would still be dominated by oncology plays. That mantra in biotech R&D recently got a big boost when Amgen presented the first cut of human data on its effort to target KRAS G12C with a small molecule drug.
Even disease-causing proteins with no known binding sites, the theory at Frontier goes, bend and create temporary pockets when they move, cracking up a window for therapeutic innovation.
“Druggable” proteins, Varma explains, are like coat hangers: They have nice corners that drugs can dock into. The other 90% of the human proteome appear, figuratively, more like a string. But in reality proteins are more dynamic than that.
“If you shake the string, you wiggle it, then you see that actually curves do form and within those curves are corners and you could imagine docking a small molecule drug into those,” Varma says. Frontier’s tech then allows them to put a covalent bind there, creating a permanent lock in a transient corner.

UC Berkeley professors Daniel Nomura and Roberto joined CEO Varma in co-founding the biotech, with chemoproteomic platforms and some basic research on cancer growth regulation in tow, respectively. That combination gives Frontier a database of hotspots (or binding pockets) in human proteins as well as a library of compounds, to be sifted through by machine learning techniques. In addition to traditional small molecule drugs, it also boasts of a “novel approach to protein degradation” — a popular and increasingly crowded field.
“It’s an alternative pathway for us to make sure that we can really go after any protein target of interest,” Varma, who left MPM a few months ago to go all in with the venture, says.
While his team of 10 — he expects to at least double that by the end of the year — is pushing as hard as they can on internal programs, he added, they are open to partnering in other cancers and therapeutic areas such as neurodegeneration.
Jim Winkler, a former CSO at protein degradation specialist Arvinas, is spearheading the degrader tech development while data science expert Johannes Hermann takes up the chief technology officer’s role. In a recent stint, Hermann headed up machine learning and advanced analytics at J&J’s Janssen.
To complete the Series A, the co-founders brought in Deerfield Management and Droia Oncology Venture alongside MPM, while DCVC Bio, RA Capital Management and other investors also participated.