A few years ago, Gevorg Grigoryan PhD ’07, then a professor at Dartmouth College, was toying with the idea of data-driven protein design for therapeutic applications. Not knowing how to implement the concept in a company, he dug up an aged syllabus from an entrepreneurship course he took during his PhD at MIT and decided to email the instructor.
He worked on the email for hours. A few sentences turned into three pages, then a few sentences again. Grigoryan finally hit send in the early morning hours.
Just 15 minutes later he received a reply from Noubar Afeyan 1987 Ph.D., CEO and co-founder of a venture capital firm Flagship Pioneer (and 2024 OneMIT graduation speaker).
This eventually prompted Grigoryan, Afeyan and others to form together Generate:Biomedicinewhere Grigoryan currently serves as Chief Technology Officer.
“Success depends on who evaluates you,” says Grigoryan. “There is no right path – the best path for you is the one that works for you.”
Generalizing principles and improving lives
Generate:Biomedicines is the culmination of decades of progress in machine learning, biological engineering and medicine. Until recently, de novo protein design was extremely labor-intensive, requiring months or years of computational methods and experiments.
“Now we can just push a button and have the generative model throw out a new protein with almost perfect probability that it will actually work. It will fold. It will have the structure that you intend,” Grigoryan says. “I think we’ve discovered these generalizable principles about how to approach understanding complex systems, and I think that will continue to work.”
From the outset, the obvious application for his work was drug discovery. Grigoryan says one reason he left academia—at least for now—is the resources available for this groundbreaking work.
“Our space has a pretty exciting and noble reason for existing,” he says. “We want to improve people’s lives.”
Mixing disciplines
STEM majors are becoming increasingly popular, but when Grigoryan was an undergraduate, there was little to no infrastructure for such education.
“A new intersection of physics, biology and computational science was emerging,” Grigoryan recalls. “It wasn’t like there was a solid discipline at the intersection of these things, but I felt like there could be one and maybe I could be a part of creating it.”
He studied biochemistry and computer science, which greatly confused his advisors in each program. It was so unusual that there were no guidelines as to which group he should go with at graduation.
I’m going to Cambridge
Grigoryan admits that his decision to study at the Department of Biology at MIT was not systematic.
“I thought, ‘MIT sounds great—strong faculty, good technical school, good city. I’ll think of something,’” he says. “I can’t emphasize enough how important and formative my years at MIT were to who I ultimately became as a scientist.”
He worked with Amy Keatingthen a junior researcher, now head of the Department of Biology, working on modeling protein-protein interactions. His work covered physics, mathematics, chemistry, and biology. He was still a few years away from starting his doctoral studies in computational biology and systems biology, but the emerging field was considered crucial.
Keating remains an advisor and confidant to this day. Grigoryan also praises her commitment to mentoring while balancing the demands of a faculty position—fundraising, running a research lab, and teaching.
“It’s hard to find the time to really mentor and help students grow, but Amy was someone who took that very seriously and was very intentional,” Grigoryan says. “We spent a lot of time talking through ideas and doing science. It’s hard to overstate the impact that mentoring can have.”
Grigoryan then completed a postdoctoral fellowship at the University of Pennsylvania. William “Bill” DeGradocontinuing to focus on protein design while gaining more experience with experimental approaches and learning other ways of thinking about proteins.
Just studying them allowed DeGrado to intuitively understand molecules—predicting their functionality or mutations that would disrupt that functionality. Its predictive power exceeded that of computer modeling at the time.
Grigoryan began to wonder: Could computational models leverage past observations to be at least as predictive as someone who had spent a lot of time considering and observing the structure and function of these molecules?
Grigoryan then went to Dartmouth as a computer science faculty member, organizing biology and chemistry meetings to investigate the issue.
Balance between industry and academia
Much of science is based on trial and error, but Grigoryan showed early on that accurately predicting proteins and how they will bind, bind and behave does not require starting from first principles. The models became more exact by solving more structures and taking more binding measurements.
Grigoryan admits that it was the leaders of Flagship Pioneering who were initially full of faith in the possible applications of the concept – more bullish than Grigoryan himself.
He spent four years dividing his time between Dartmouth and Cambridge until he finally decided to leave academia entirely.
“It was inevitable because I was so in love with what we had created at Generate,” he says. “It was so exciting for me to see that idea come to fruition.”
Stop or grow
Grigoryan says the most crucial thing for the company is to scale at the right time, striking a balance between “strike while the iron is hot” and taking into account company, technology and market readiness.
However, even successful growth creates its own challenges.
When there are fewer than twenty people working in a company, adapting strategy across the company is uncomplicated: everyone can be in the room. But growth—say, expanding your workforce to 200 employees—requires more thoughtful communication and balancing flexibility while maintaining your company’s culture and identity.
“Growth is hard,” he says. “And it takes a lot of intentional effort, time and energy to ensure a clear culture that allows a team to thrive.”
Grigorian’s time in academia was invaluable as he learned that “it’s all about people” – but academia and industry require different approaches.
“Being PI [principal investigator] is to create a path for each of the interns in which they are essentially independent scientists,” he says. “In a company, you are inherently bound by a set of common goals, and you have to value your work by the degree of synergy it has with others, not by what you can do alone.”