A fusion of this year’s buzziest biotech trends — artificial intelligence and obesity drugs — is the focus of a new startup backed with $53.5 million from biotech investors Versant Ventures and DCVC Bio, Endpoints News has learned.
The startup, called Fable Therapeutics, has been operating in stealth for nearly three years. It’s based on research from co-founder and chief technology officer Philip Kim, a scientist at the University of Toronto and an early pioneer in the now booming field of using AI to create proteins never before seen in nature.
Fable is building on Kim’s diffusion models — the type of AI used to generate fanciful images from text prompts — to make protein therapies, mainly antibodies but also peptides to potentially create new obesity medicines that go far beyond what existing GLP-1 agonists like Ozempic and Zepbound, and even their successors, can do.
Fable CEO Geoff MacKay — who joined the startup just months before his prior company, gene therapy developer Avrobio, halted all drug development last summer — told Endpoints that he thinks AI will give Fable an “unfair advantage” in creating a “generation three of this exploding obesity market.”
The startup is developing drugs that go “beyond basic weight loss” and help patients improve their basal metabolic rate, bone density, insulin sensitivity and strength, MacKay said. He thinks such drugs could be used in combination with existing or improved weight loss drugs to address the comorbidities of obesity and the unwanted effects of GLP-1 drugs, such as loss of muscle.
MacKay wouldn’t name any of the targets that Fable is focused on or say when its drugs will enter clinical trials, but he expects to have development candidates for multiple programs in 2025. The startup also has four small research partnerships with pharma companies, but he couldn’t disclose the details.
AI antibody boom
Kim is a highly-cited biologist and machine learning scientist, but he admits that he was initially skeptical about using AI to make proteins until one of his students showed him early results in 2017. It was “mind-blowing,” he said. “The strongest zealot is the converted skeptic.”
Since then, he’s racked up a long list of firsts at the intersection of AI and proteins. He was the first to publish generative models for protein backbones in 2017, the first to make a graph neural network for protein design in 2019, and the first to release a validated protein backbone diffusion model in 2022.
AI for predicting protein structures and designing new ones was the subject of a Nobel Prize awarded to Kim’s competitors earlier this month. It’s also attracted immense investment, including a more than $1 billion launch for Xaira Therapeutics, a startup co-founded by Nobel winner David Baker; nearly $750 million across several funding rounds for Generate:Biomedicines; and a $142 million seed round for EvolutionaryScale.
Fable’s comparatively modest $10 million seed round and $43.5 million Series A are intended to fund the company into 2026, MacKay said. The startup has a dozen machine learning scientists in Toronto and another dozen protein researchers in Boston.
The specifics of how Fable is improving Kim’s technology, and how the startup will use it, are still under wraps. But like other startups in the field, MacKay has a bold vision for how AI will change antibody drug discovery and said it will help the company optimize multiple properties of proteins.
“The first generation of the platform was pretty mixed, but already by the time we got to generation two, you could produce out-of-the-box hits,” he said.
Editor’s note: This story was updated to indicate that life science fund DCVC Bio was an investor, not the tech fund DCVC.