Flagship Pioneering debuts new startup using AI to design disease-fighting proteins

After raising $1.1 billion earlier this year to back a new slate of biotechs, Flagship Pioneering has raised the curtain on an artificial intelligence company focused on discovering and designing a wide range of new drugs.

Generate Biomedicines will use a machine learning platform, developed under-the-radar over the past three years, to help parse the structure of thousands of human proteins and create novel protein sequences that could form the basis of new treatments.

The company also plans to use its generative chemistry and biology systems to invent new therapies taking the form of AI-designed antibodies, peptides, enzymes and cytokines, as well as developing new gene-editing technologies.

"Breakthroughs in machine learning algorithms, the exponential growth in computing power, and the acceleration and democratization of DNA sequencing and DNA synthesis are allowing us to learn from biology at unprecedented scale," said Flagship’s founder and CEO, Noubar Afeyan. 

"Protein design is not a new idea, but it has been frustrated by the limitations of previous technologies,” Afeyan added. “Generate Biomedicines was formed to move biomedicine past its dependence on existing discovery methods and develop a new machine learning technology that can generate biologic drugs for potentially any target, in order to treat previously intractable diseases."

RELATED: Flagship raises $1.1B to create biotechs for post-pandemic world

The company was co-founded in 2018—by Flagship’s Avak Kahvejian, Geoffrey von Maltzahn, Molly Gibson and Gevorg Grigoryan, who also maintains a lab at Dartmouth University—after two Flagship Labs projects focused on creating drugs out of proteins were merged together. 

"Generate began with the question, 'What if we could generate novel protein therapeutics using new computational tools without having to discover them through trial and error?'" said von Maltzahn, who will serve as co-CEO along with Kahvejian.

"In much the same way that the patterns found in large libraries of songs, texts or photographs have been used to create AI-generated music, language and faces, we have shown that patterns in protein sequences and structures can be broadly applied to generate novel biomedicines,” he said.

RELATED: Flagship Pioneering combines Torque and Cogen Immune Medicine to rev up R&D

Generate said it has begun constructing a portfolio of candidates that it plans to develop itself as well as put up for partnerships with other biopharma companies. 

This includes potential therapies for COVID-19, including antibodies and peptides that target the novel coronavirus’ spike protein, which is used to enter and infect human cells. The company said that, over a period of 17 days earlier this year, it was able to generate 100 potential antibodies and narrow them down to those with the best chance of binding to the virus and slowing its spread.

That project will continue its work in collaboration with the Coronavirus Immunotherapy Consortium, an initiative led by the La Jolla Institute for Immunology and supported by the COVID-19 Therapeutics Accelerator, launched earlier this year by the Bill & Melinda Gates Foundation, the Wellcome Trust and Mastercard.