Genomic discovery on the road less traveled

“There are a multitude of reasons why what this company is doing is revolutionary,” says Keolu Fox, a scientist of Native Hawaiian descent at the University of California, San Diego, whose research is focused on developing and applying new technologies in genomics. “First and foremost, those that we recruit for 95% of clinical trials in America are of Western European ancestry. And the same goes for genomic studies. So the type of data that Variant Bio is generating is more valuable than ever.”

SoftBank Investment Advisers’ John Xue, a Variant Bio board member, adds: “They’re solving the biggest bottleneck that exists in drug discovery.” While the human genome has 20,000 genes, approved medicines address about 700 targets, he says. “Humanity has a dearth of novel targets for diseases with unmet medical need. The potential value unlocked is enormous.”

Meanwhile, Variant Bio has also developed an AI-powered analytics platform. Its primary purpose was to be a tool that would help the company extract insights from the data it collected. But the platform has recently drawn the attention of outside companies, including international drugmaker Novo Nordisk, which has announced that it will pay Variant Bio up to $50 million to help it identify specific targets using the platform, such as proteins or genes that could be modified to treat disease. 

“Drug discovery is still our North Star,” says Castel, who is chief genomics officer. “But we’re open to exploring other opportunities, such as letting partner companies use the platform, in collaborations like Novo, that will help us scale.”

Zeroing in on differentiated data

As a child growing up in Toronto, Castel was fascinated by exploration and technology. He cherished old-school “Choose Your Own Adventure” text-based games — interactive fiction where the player’s choices determine the story’s path — and was especially drawn to adaptations that offered futuristic scenarios. He loved nature and the outdoors, but was also into fixing old computers.

It was a big, bright poster hanging on the back wall of a high school biology classroom that proved most thrilling. Titled “Annotation of the Celera human genome assembly,” it showcased one of mankind's greatest accomplishments: mapping the human genome.

“It struck me: Oh, wow, this is like programming — but for life,” he says. “Something just clicked. I knew from that moment on that I wanted to be a geneticist and use computers to analyze genetic data.”

Over cocktails in a New York City bar in 2018, after earning a Ph.D. from Cold Spring Harbor Laboratory and completing postdoctoral computational genetics training at Columbia University, Castel and fellow geneticist Kaja Wasik conceived of Variant Bio.

Their goal was to leverage the enormous power of human genetic diversity to transform drug discovery and development. Identification of changes in DNA called mutations or “variants” can point to ideas for novel therapeutics for common diseases across humanity. (Wasik, an adjunct professor at the J. Craig Venter Institute and a visiting scholar at Princeton University, has since left Variant Bio to pursue her passion for wildlife conservation.)

“Everyone who is developing drugs wants genetic evidence to back those drugs, because they are much more likely to work when you take them into the clinic,” says Variant Bio CEO Andrew Farnum.

They knew that the study of populations with high incidence of specific medical traits and unique genetic architecture could simplify the search for “targets” — generally, molecules or proteins — that are involved in a disease process. 

Working in remote locales, Variant Bio’s team collects DNA samples for later sequencing. It also conducts broad and deep phenotyping — measuring each person’s biochemical and immune markers, as well as blood pressure, heart rate, activity level, height, weight, lifestyle, health history, and characteristics related to specific conditions.

By collecting and understanding this unique data at a granular level, Variant Bio can gain insights that its larger competitors may miss. In the AI era, what matters is not just the quantity of information, but also its quality and consistency — and the tool that manages it.

“The key is having really high-quality data,” says Castel.

The first order of business for Variant Bio was to show that it could actually work effectively and cooperatively with Indigenous and other underrepresented populations, as well as manage the complexities of gathering data in highly varied environments. In recent years, the startup has shown it can handle the task, while pioneering a widely praised model of collaboration.

“Then we had to prove that that data was actually useful,” says Farnum. “That’s where we are now. We have a number of novel insights through the data that we’ve gathered around the world and the platform that we’ve built. And we’re working to turn those insights into drugs that we can take into clinical testing.” 

Even small genetic changes can have large and far-reaching effects. So far, the team has found genetic variants linked to fibrosis, lupus, and muscle wasting, and has developed potential drugs — dubbed VB-8, VB-11, and VB-12, respectively — for treatments. 

“It’s a journey of validating our approach, step by step,” says Farnum.

Genomic research across the globe

To be sure, plenty of giant genomic and health databases already exist. The UK Biobank contains information from 500,000 participants. In the United States, the National Institutes of Health’s “All of Us” research program has nearly 250,000 genome sequences. 

But they largely focus on a narrow slice of humanity, such as people of Western European ancestry. This yields similar genetic data, which limits any novel insights, says Farnum.

“Because everyone is looking at the same few datasets, everyone is finding the same targets,” says Farnum, who came to Variant Bio in 2020 from the Gates Foundation, where he ran its $2.5 billion strategic investment arm. “We have data others don’t have, which allows us to find genetic evidence for targets that others don’t know about.”

This genetic evidence can be more readily discovered by focusing on historically isolated populations. Centuries of endogamy, or marriages within the limits of the local community, can concentrate DNA variants, which can both increase and decrease a person’s risk of developing diseases.

While a population’s particular variants may be unique, the diseases Variant Bio is researching are not. The underlying mechanisms are the same, no matter where the patient lives. 

One of the most successful examples came out of West Africa. In that region, some people have certain variations in the PCSK9 gene that are linked to lower LDL or “bad” cholesterol levels. Based on that finding, pharmaceutical companies developed drugs, known as PCSK9 inhibitors, that block the action of the gene.

“It’s an African variant,” says Farnum. “But if you or I took a PCSK9-targeting drug, it would work on us, too, because we’re all human.”

Variant Bio’s team seeks to work with populations with relatively less genetic diversity due to a phenomenon, known as a population bottleneck, during those groups’ early histories. For example, as Polynesians explored the South Pacific islands, small subsets of the original navigators settled into new homes. As the genetic pool decreased, certain genes became more concentrated.  

The company is also working with people with large genetic diversity, such as those descended from early inhabitants of the Great Lakes region of East Africa — the cradle of human evolution over 300,000 years — who have a wide array of variants. 

The company has other criteria, too: The diseases suffered by these populations must be common, lack effective treatments, and fall within Variant Bio’s focus of therapeutics.

Once on the ground, its teams collect thousands of different biological “data points” for each participant.

“The more information we have on a sample, the more confident we can be when we see something,” says Stacey Drabic, Variant Bio’s senior vice president of biology.

Discovering the power of the platform

The data is integrated with “VB-Inference,” the company’s analytics platform that uses statistical genetics and AI. It combines different levels of biological information to create a more complete picture of complex molecular processes that underlie health conditions. This points the way to possible therapies.

An AI-based tool called “Inference Chat” helps with database exploration. “Our approach lets you specifically ask: Is this gene — this target, from a therapeutic perspective — driving the disease?” says Castel, who built the platform.

The platform’s broader potential was largely overlooked at first, as the company was hyperfocused on the goal of finding future medicines. 

But in discussions with potential business partners, the team came to understand it had developed a powerful and highly differentiated technology. “I had my eyes opened up,” he says. “There are a lot of folks who are interested in our ability to analyze this data in a way that comes up with meaningful findings.” That realization led to the idea of monetizing the platform itself, with the Novo Nordisk partnership being the first tangible step in that direction.

For Variant Bio’s founders, this partial pivot served as a critical lesson in entrepreneurship. “It’s important to be responsive and say, ‘This is not what we initially conceived, but we have a new opportunity that we have to follow,’” Castel says. “The inability to change in the face of new data or a new understanding of a business is one of the worst things that can happen to a startup.”

Ethics as a North Star

From the start, Variant Bio was determined not to repeat the mistakes other companies have made in working with Indigenous and other underrepresented groups. Many have a long history of colonial exploitation and are wary of collaborating with scientists and companies from abroad.

When Variant Bio is interested in making contact with a community, it’s often the company’s director of ethics and engagement, Sarah LeBaron von Baeyer, a cultural anthropologist with a Ph.D. from Yale, who takes the first steps. Attuned to cultural sensitivities, she seeks out local experts for collaboration. “They often have longstanding ties to the community, and so they are a bridge for us,” says LeBaron von Baeyer. “We’re not just going in cold, saying ‘Hey, we’re this American company and we want to talk to you.’” 

Variant Bio has also pioneered a model that ensures any benefits flow in both directions. It shares its proceeds and medical insights with participating communities. Guided by an ethics advisory board, the company’s “benefit sharing” effort has already distributed more than $1 million to communities to support health, education, environmental, and cultural initiatives. And the company has committed to sharing 4% of its future revenue with participants. Furthermore, through its Affordable Medicines Pledge, Variant Bio aims to make any medicines or other products it develops available at an affordable price to the communities whose data first led to those discoveries.

“The benefit-sharing program builds trust within communities who want to participate in research but have historically been distrustful,” says Fox, the UC San Diego scientist, who also serves as a senior advisor to Variant Bio. “People aren’t just subjects — they’re partners in the research. I hope that that becomes something other companies want to incorporate and emulate.”

As it expanded its research to more and more remote corners of the world, Variant Bio has run into unexpected challenges.

In Madagascar, cars broke down and got stuck in muddy, flooded roads. Some villages lacked electricity. The team met participants under the canopy of a large tree. Political unrest in the French Pacific territory of New Caledonia triggered protests, curfews, and fires near the team’s biochemistry lab last year. “We were very worried that participants’ samples might get damaged,” says LeBaron von Baeyer. 

Future treatments  

Variant Bio’s approach could transform modern medicine by creating therapeutics that treat the underlying cause rather than the symptoms of disease. 

“What we’re always looking for is what is the main mechanism? Can we find it?” says Drabic. The genetic data may suggest an intriguing target — involving an underappreciated molecule, protein, or pathway — that hasn’t been revealed before. 

To decide where to focus, Variant Bio’s drug development team takes a hard look at the VB-Inference platform’s constellation of information to help it understand how to turn genetic insights into therapies. “We prioritize on the basis of how strong the data is, both from the genetics and from whatever else we can find,” says David Moller, Variant Bio’s former chief scientific officer and current board member, who is a physician scientist with lengthy experience in pharma and biotech. “We put two and two together, and we come up with a hypothesis.”