Tomorrow’s chemistry, today: Bringing materials science into the 21st century
Why ArcTern invested in Kebotix’s Series A
ArcTern Ventures is pleased to announce our investment into Series A of Kebotix, a Cambridge-based technology platform company for new chemicals and materials dedicated to improving public health, championing sustainability and eliminating the production of hazardous substances through Green Chemistry solutions. Kebotix partners with the private and public sector in harnessing the power of its breakthrough platform that combines artificial intelligence and robotic automation to discover chemicals and materials significantly faster and more affordably.
ArcTern participated in the Series A led by Denmark-based Novo Holdings, a leading international life science investor, and joined a group of existing investors including One Way Ventures, Flybridge Ventures, Embark Ventures and Propagator Ventures.
Many major scientific discoveries, whose usage and benefits are ubiquitous today, have been results of ‘happy accidents’.
Serendipity has played a crucial role in science, with majority of the most important and revolutionary discoveries in biology and medicine having an element of ‘lucky accident’. Many of the incredible discoveries that have benefitted the world came about when scientists found something they weren’t looking for. As an example, a soil sample near a golf course outside of Tokyo unearthed a bacterium producing compounds that impede the activity of nematode worms. The compounds were then developed into a drug that can prevent or treat human parasitic diseases that would otherwise cause blindness in more than 250 million people each year. This earned Satoshi Ōmura and William Campbell the 2015 Nobel Prize in Physiology or Medicine (shared with Youyou Tu, who discovered a treatment for malaria). The myriad of similarly serendipitous examples also includes penicillin, insulin, safety glass used in windshields, vulcanized rubber, Teflon, superglue, and the artificial sweetener “Sweet’N Low”.
On the other hand, the hypothesis-driven scientific method, evolved over the last four centuries, had been the best process humanity has devised to date to learn about the universe around us. This process of science can be crudely simplified as follows:
- State a hypothesis, an educated guess about how the natural universe works, typically a testable extension of previously understood knowledge
- Test the hypothesis via an experiment to see if a guess (hypothesis) is correct
- Refine the hypothesis based on experimental results
The scientific method has certainly contributed to an exponential growth in our society’s knowledge base and development of new technologies that have enabled more diverse and efficient economic activity, growing population and higher quality of life.
The pace of technological innovation has generated unprecedented demand for new materials, and material innovation has enabled new technologies not previously possible. At the same time, it has become paramount for us as a society to accelerate the discovery and commercialization of green chemistries that can help mitigate major environmental impacts such as GHG emissions from air conditioning and proliferation of plastic waste across landfills and oceans.
Unfortunately, the technology cycle time (TCT) — median age of the patents cited on the front page of a patent document — for U.S. chemical patents is flat. This indicates that the chemical industry’s rate of innovation has decreased, despite the consistent R&D spend in tens of billions of dollars annually, with just the top 20 global chemical companies accounting for > $11B.
Breaking the innovation stagnation cycle, Kebotix offers a “closed loop innovation” that combines artificial intelligence and robotic automation to discover chemicals and materials significantly faster and more affordably.
- By using inverse design technology instead of the hypothesis-driven approach, Kebotix algorithms can look for optimal molecular combinations that meet specific parameters (e.g. toxicity) to generate highest potential ‘lead candidates’ not just from a pool of ~7 million known chemicals and materials, but from an intrapolated universe of 10⁶⁰ potential chemically active combinations ‘beyond the bounds of human intuition’
- The company’s platform is differentiated for its ability to generate own data, by using several leading machine learning techniques to simulate chemical properties and reactions before physical testing, as well as via autonomous, self-driving lab automation to generate new experimental data
In addition to generation of synthesizable new chemicals and materials, the Kebotix team has already started to demonstrate the capabilities of its platform in accelerating the chemicals and advanced materials R&D pipeline across:
- Lead optimization, through properties prediction and high throughput virtual screening of thousands of potential candidates to shortlist few viable leads for testing in up to 10x less time (e.g. in an executed paid partnership to identify top candidates for a chemical alternative with significantly less GHG impact, in only 3 months yielded multiple ‘hits’, which otherwise would have required the corporate partner to conduct 9X more physical testing over a timeline that could stretch to several years)
- Optimization of experimental parameters to enable finding of optimum experimental process ~5–8X faster
- Process optimization to enable more optimal chemical production at manufacturing scale (e.g. currently successfully delivering on a paid partnership with an oil and gas major to innovate a cost-effective way of bio-polymer production using the Kebotix synthetic route prediction tool)
While Kebotix’s closed loop model is new within material science, it has already been successfully advanced in pharma. The last few years have seen over $5B in VC funding towards artificial intelligence for drug development, most of it focused on drug design, with leading startups executing multiple joint venture partnerships with major pharmaceutical companies potentially valued in the billions. Earlier this year, a drug molecule “invented” by artificial intelligence will be used in human trials in a world first for machine learning in medicine. The drug, to be used to treat patients who have obsessive-compulsive disorder, was created by British startup Exscientia and Japanese pharmaceutical firm Sumitomo Dainippon Pharma, taking only 12 months to get to trial relative to the typical 5-year drug development period. ArcTern is excited to support the Kebotix team in continuing to execute on a similar business model.
