In 2013, I was working on my Ph.D. in Bioengineering at the University of Utah. I was part of a team that was searching for potential treatments for cerebral cavernous malformation (CCM), a rare, genetic disease that causes clumps of leaky blood vessels in the brain. CCM can lead to seizures, speech or vision problems, and even brain hemorrhages. While there are three genes known to cause CCM, there was (and still is) no approved drug to treat it.
‍
While at the University of Utah I began to have a first-hand appreciation for how traditional, reductionist approaches to drug discovery fail patients. For some time we tested one drug for CCM that was based on historical knowledge and understanding of the disease. This drug ultimately made the symptoms worse in animals. As someone maybe newer and naive to the field I was incredibly frustrated; how is this the best we can do? Is there no more efficient, more effective way to discover potential treatments than relying on the limits of our own intelligence? This idea that we, as humans, know so little about complex biology really bothered me. And so, our team began to use a newer tool to explore the biology of the disease: machine learning.
‍
Our team used open-source machine learning software to scan cellular images and probe the effects of more than 2,000 drug compounds, looking for ones that improved the function of blood vessel cells carrying the defunct genes. These machine learning algorithms ultimately led us to a chemical that reduced the blood vessels by 50% in animal tests. This was incredibly validating and gave me an idea.
‍
I began thinking less about becoming a surgeon — my original, chosen path — and more about the potential, exponential impact we could have by applying machine learning to cell images and generating massive datasets that would reveal new biological interactions, potential drugs, and ultimately, industrialize a process that is ripe for innovation. I thought: I could be a surgeon and impact maybe 100 or so people a year OR I could start Recursion and someday, hopefully, impact hundreds of millions of lives over years.
‍
And so, along with my co-founders Dr. Dean Li (then head of the lab I worked in, now the SVP of Translational Discovery at Merck) and Blake Borgensen (our original CTO and today an ardent advocate for AI ethics), we founded Recursion in 2013. The CCM program became our first clinical-stage program and remains one of our lead programs today. Fast forward to today: six years in business and the progress we have made in that time is unmatched.
‍
Drug discovery is hard. We don’t pretend to make it easy. But by embracing the wave of advanced technologies and thinking introduced by the fourth industrial revolution — machine learning, high-throughput sequencing, CRISPR editing, and more — we stand to bring about a new era of discovery that is more data-driven and predictive than ever before. We can, finally, build a map of human biology, empowering us to decode biology and radically improve people’s lives.
‍
It is that ambitious, bold vision that has gotten me up early and excited every single day since the first day I helped start this company. The future is bright and the path seems clear; digital biology is here to stay, and I am proud for Recursion to be a driver of it.

— Chris Gibson