When biotech startups emerge from stealth, executives typically talk about new molecules in the pipeline and novel approaches they take to treating a disease. The founders of Generate Biomedicines want to discuss none of those things, yet they contend their technology will upend the way protein drugs are discovered and developed.
Cambridge, MA-based Generate has been plugging away at its research within the labs of venture capital firm Flagship Pioneering for the past three years. On Thursday, the startup pulled back the curtain enough to give a glimpse of its technology, its potential applications, and the plans to use that platform to create novel drugs that treat disease in new ways.
Proteins are known to play a part in both health and disease, but less understood is protein function—the biological role of these molecules. The rules defining how a protein is made and what it does in the body are set out by Mother Nature, says Geoff von Maltzahn, Generate’s co-CEO. Generate’s technology aims to decipher those rules by analyzing “every known protein on every branch of the tree of life,” von Maltzahn says.
Generate uses machine-learning techniques to study proteins. The startup’s algorithms analyze the hundreds of millions of known proteins in search of patterns in protein sequences. Avak Kahvejian, the company’s co-CEO, says that what’s needed for the computational technology to work is for scientists to pose a question for the system to answer: What is a protein’s function? What is the best binding target? “Computers are very bad at asking questions but they’re very good at finding answers,” Kahvejian says.
Those answers come quickly—in weeks compared to the months it takes using existing technologies, Kahvejian says. More than simply understanding protein function, von Maltzahn says the analysis enables the company to make predictions about a protein, the target or targets it could hit, and the therapeutic effect it will have. The protein drugs that emerge from this analysis could come in a number of forms: a peptide, antibody, enzyme, or cytokine.
Generate isn’t the first company to apply machine-learning techniques to drug discovery. But Molly Gibson, the company’s chief innovation officer, says that earlier approaches try to understand an aspect of proteins, such as protein folding. Beyond understanding the protein’s structure, Generate aims to turn its analysis into new proteins.
“Often they’re just making small changes to what exists,” Gibson says. “We’re doing entirely new realms of sequences.”
In addition to creating novel proteins, Generate says its technology can also take known proteins and get them to work in new ways. As an example, Kahvejian points to bacterial proteins. These proteins produce a strong immune response but they are hampered as therapies for chronic indications because of the body’s limited ability to tolerate them. Kahvejian says Generate can engineer bacterial proteins to make them look like human proteins on the outside while retaining the same catalytic activity as the original protein.
Startups that emerge from Flagship begin as research projects aiming to find an answer to some question about the human condition. Examples include the plant micobe inquiry that led to agbiotech company Indigo and the red blood cell therapy research that produced Rubius Therapeutics (NASDAQ: RUBY). The venture capital firm runs multiple exploratory programs simultaneously; Generate is the product of two separate programs. One program, led by Kahvejian, aimed to use discoveries about protein structures as the basis for an algorithmic technology platform for drug discovery.
Another Flagship program led by Gibson and von Maltzahn aimed to see if the advances in machine learning in uses such as processing language and images could also apply to proteins. Flagship’s partners came to see that that both projects involved a shift in the understanding of protein structure and they combined the two research efforts, leading to Generate’s technology for analyzing proteins and creating protein drugs.
Generate’s research has produced drug candidates, but von Maltzahn declined to discuss those molecules or their disease targets. Broadly speaking, he says the company’s technology has created novel antibodies, peptides, and enzymes for as many as 50 potential applications, revealing that one of them is gene therapy. Kahvejian adds that Generate has applied its technology to COVID-19, research that has yielded antibody and peptide candidates that address multiple targets on the spike protein of the novel coronavirus. Those candidates are being developed in partnership with the Coronavirus Immunotherapy Consortium, a global research collaboration.
Like most Flagship companies, Generate’s technology is a platform that has multiple potential applications, Kahvejian says. The company’s near-term focus is to further develop that technology for applications in human health and select which programs to advance toward human testing and which ones to develop in partnership with larger companies. Partnering discussions are already underway, von Maltzahn says.
Funding is yet another thing that Generate’s founders are reluctant to discuss. Von Maltzahn says Generate has been backed by $50 million from Flagship and the startup is not looking for additional capital right now. Kahvejian says that as Generate continues to grow, the company will be looking to raise money from other investors some time next year.
|Want more Xconomy content? Subscribe today for free newsletters, event and webinar alerts, whitepapers, podcasts, and more.|