Osmo, a machine olfaction startup, announced the receipt of a $3.5 million grant from the Bill & Melinda Gates Foundation to advance the company's AI-enabled scent platform for discovering and producing compounds that repel, attract, or destroy disease-carrying insects to improve animal and human health. The grant is in addition to a $5 million equity investment in Osmo that the foundation made at the company's January 2023 launch.
The World Health Organization (WHO) estimates that disease-carrying insects such as mosquitoes cause millions of deaths worldwide each year. Because insects rely heavily on their olfactory senses for navigation and locating potential hosts, scent is the most direct way to steer disease-carrying insects away from humans. Developing compounds with specific scents that repel or deter insects can effectively disrupt their attraction to human hosts, minimizing disease transmission and providing a targeted, efficient approach to insect control.
"New scent molecules that more effectively steer disease-carrying insects from human contact have the potential to save millions of lives," said Osmo CEO Alex Wiltschko. "In the vast space of billions of potential molecules, only a few thousand have ever been screened for such capabilities. With generous support from the Gates Foundation, we're using our AI Scent Platform to analyze this vast chemical space and discover novel agents that modify the behavior of insects to prevent disease and are effective, safe, and affordable for both human and animal health globally."
This grant follows a proof-of-concept pilot previously funded by the foundation and published as a research paper in late 2022. In the pilot, the research team trained a computational model on history's largest mosquito repellency dataset to date. Evaluating the model experimentally on ~400 chemically diverse repellency molecules, the team identified eight molecules that were more repellent than that of the widely used DEET and Picaridin.
In the current project, Osmo will build on the proof-of-concept pilot with the primary aims of:
- scaling up the previous research at least tenfold by incorporating more data and more tests on more compounds;
- discovering through machine learning promising candidate molecules that are novel, inexpensive, and chemically diverse;
- developing predictive models to incorporate real-world constraints associated with the candidate molecules, including cost, spatial range, biodegradability, toxicity, and human odor perception;
- synthesizing, testing, and optimizing novel candidate repellents for advancement to human trials and ecological impact assessments; and
- leveraging this repellent model to demonstrate the efficacy of ML-driven discovery of novel mosquito attractants that outperform existing lures.
Osmo's insect-control platform design leans heavily into recent advances in machine learning and generative AI, which enable – in a matter of seconds – the evaluation of billions of molecules for their potential effectiveness and safety.
"New machine learning approaches have major potential to speed up the discovery and design of improved mosquito repellents and attractants," said Meg Younger, Assistant Professor in Boston University's Department of Biology. "Osmo's model shows great promise, and I'm looking forward to tracking the team's progress in the coming years."
Osmo's insect control work is part of the company's mission to give computers a sense of smell to improve the health and well-being of human life. Core to this mission is a "map of odor" built by the Osmo team to predict what a molecule smells like from its structure.
"Osmo's science revealed a surprising link between insect and human olfaction, with our map of odor predicting how molecules smell to humans as well as insects," said Wiltschko. "Our mission to digitize olfaction will have many potential paths to make the world healthier and happier. All of us at Osmo are particularly excited about using our map to design totally new molecules to stop the spread of insect-borne disease."
For more information: