Training machine learning (ML) and artificial intelligence (AI) algorithms requires quality data. In the pharmaceutical sector, obtaining this data can be tough because most of the knowledge about drugs is contained in scientific publications in an unstructured text format.
Enter Edmonton-based DrugBank, which uses a “hybrid approach” that involves AI, natural language processing (NLP), and scientific experts to structure this data and make it usable for pharma companies and academic researchers. With over 12 million annual users, 300-plus open source projects, more than 26,000 academic citations, and a customer base that includes 13 of the world’s top 20 pharmaceutical companies, DrugBank claims to be “the world’s largest online database of structured drug and drug target information.”
“Eventually, most of drug discovery … and most of medicine will be powered by AI.”
-Mike Wilson, DrugBank
To date, the University of Alberta spinout has achieved most of its growth—including its 200-plus paying customers—organically. Now, with $9 million CAD in seed financing led by Montréal-based Brightspark Ventures, and supported by a group of strategic investors from the precision medicine and pharma spaces, DrugBank is ready to grow its team and expand its AI capabilities.
In an interview with BetaKit, DrugBank co-founder and CEO Mike Wilson said he sees “a lot of room” for the startup to grow in the pharma industry, where it serves a mix of giants like Merck and smaller firms that are leveraging AI in the drug discovery and precision medicine spaces.
“Those are two areas that are very data dependent … and we see them growing a lot,” said Wilson. “Eventually, most of drug discovery will be powered by AI and most of medicine will be powered by AI, and it will be using this data in much, much more advanced ways.”
DrugBank’s all-equity March seed round was led by Brightspark, with participation from Montréal-based precision health-focused investment fund Amplitude Ventures and Belgium’s Theodorus Investment Funds—which has “a strong network” of connections to pharma companies operating in the country. The round also saw support from existing investors AccelerateFund and Milad Alucozai from BoxOne Ventures, who led DrugBank’s 2020 pre-seed financing. The raise brings DrugBank’s total funding to $10.6 million CAD.
DrugBank’s origins date back to 2006, when it was launched as a University of Alberta project in Dr. David Wishart’s lab focused on helping academic researchers gain access to structured info about drugs. In 2011, the initiative became a part of the federally-funded, Edmonton-based Metabolomics Innovation Centre.
RELATED: Zymeworks co-founder, former CEO Ali Tehrani joins Amplitude Ventures as Venture Partners
Recognizing the project’s potential following “a lot of inbound interest from companies,” Wilson and CTO Craig Knox—a pair of researchers who had previously worked on the initiative—decided to commercialize DrugBank in 2015, nabbing their first customer in less than a month.
“The demand was already there, so once we commercialized, we had the world’s top pharmaceutical companies knocking at our door,” said Wilson.
Given DrugBank’s progress to date, Brightspark Principal Eleonore Jarry told BetaKit that the startup is “uniquely positioned to win” in this space.
“The company has extensively researched and developed proprietary technology, over 15 years in the making,” said Jarry, who cited DrugBank’s strong reputation in the pharma sector and organic growth as two of the factors that drew Brightspark to invest.
RELATED: Deep Genomics secures $226 million CAd led by SoftBank as it scales AI drug discovery platform
According to Jarry, Wilson and Knox bring “a unique knowledge mix of computing and biology” to the table. As part of the startup’s seed round, Jarry is joining DrugBank’s board, along with Theodorus Principal Théo Risopoulos.
DrugBank is a member of Edmonton’s growing AI ecosystem, which includes support organization Amii, as well as growing startups like Samdesk and PainWorth.
Investment in AI-powered drug discovery is on the rise. Within the Canadian tech sector, there are a number of different companies working at the intersection of AI and drug discovery, including Vancouver-based AbCellera and Toronto’s Deep Genomics. DrugBank aims to become the “critical infrastructure” for firms operating in this space.
According to Knox, the complexity of biomedical data—not to mention the amount of it—is increasing. “With over two million scientific journals being published each year, the rate at which biomedical data is growing and the speed at which it becomes out of date makes it difficult to rely on and even harder to work with,” he said.
RELATED: AbCellera’s antibody treatment becomes first to receive emergency FDA approval for kids under 12
Amid this environment, DrugBank’s goal, says Wilson, is to enable all of that information “to be used to its full potential” by customers of all sizes.
“The saying goes ‘during a gold rush, sell shovels,’” said Jarry. “DrugBank is the shovel in this new era of medicine, providing the structured drug data necessary for both incumbents and startups to develop their own AI solutions.”
DrugBank’s goal is to become a standard source of biomedical data used by every pharma company. To get there, the startup intends to use its seed financing to invest heavily in research and development (R&D), expand its AI and NLP capabilities, and double its 50-person team over the next year.
Wilson sees plenty of room to grow by catering to this space, and so do the company’s investors. Amplitude co-founder and partner Jean-Francois Pariseau told BetaKit he believes DrugBank has “a massive opportunity in front of it.”
“The adoption of AI and [ML] across the industry is just beginning—99 percent of companies have been involved in a project that has AI as part of the project, but this is the beginning of a deeper integration of AI tools in the drug development process,” said Pariseau. “These tools all need rich, structured data to build their algorithms.”
Feature image of Craig Knox and Mike Wilson, courtesy DrugBank.
