Allozymes’ ingenious approach to rapidly testing tens of millions of chemical reactions in biological media is proving to be not only a useful service, but also the basis for a unique and helpful dataset. And where there is a data set, there is artificial intelligence, and where there is artificial intelligence, there are investors. The company just raised $15 million in Series A funding to grow its business from a useful service to a world-class resource.
We first wrote about the biotech startup in 2021, when it was taking its first steps: “Back then, we were less than five people and our first lab was a thousand square feet,” recalls CEO and founder Peyman Salehian.
The company has grown to 32 people in the U.S., Europe and Singapore and has 15 times more lab space, which it has used to speed up its already exponentially faster enzyme screening technique.
The company’s core technology has not modified since 2021, and you possibly can read its detailed description in our original article. The upshot, nevertheless, is that enzymes, or chains of amino acids that perform specific tasks in biological systems, have so far been slightly difficult to find or invent. This is because of the enormous variety of variations: a molecule will be lots of of acids long, with 20 to select from for each position, and each permutation potentially produces a completely different effect. You quickly gain access to billions of possibilities!
Using traditional methods, these differences will be tested at a rate of several hundred per day in a reasonable laboratory space, but Allozymes uses a method in which tens of millions of enzymes will be tested per day by packaging them into tiny droplets and passing them through a special microfluidic system. You can think of it as a conveyor belt with a camera above it, scanning each item because it approaches and mechanically sorting it into different bins.
Droplets containing enzyme variants are evaluated and, if crucial, directed to the microfluidic system. Image credits: Allozymes
These enzymes can include the whole lot needed in the biotechnology and chemical industries: if you would like to convert raw materials into specific desired molecules or vice versa, or perform many other basic processes, enzymes are the right solution. Finding low cost and effective ones is rarely easy, and until recently the entire industry was testing about a million possibilities a 12 months – a number Allozymes goals to multiply by greater than a thousandfold, reaching its goal of seven billion variants by 2024.
“[In 2021] we were just building the machines, but now they work very well and we check up to 20 million enzyme variants a day,” Salehian said.
This process has already attracted clients from a wide selection of industries, some of which Allozymes cannot disclose due to NDAs, but others that have been documented in case studies:
- Phytoene is an enzyme that happens naturally in tomatoes and is often harvested in small amounts from the skins of tens of millions of tomatoes. Allozymes have found a way to make the same chemical in a bioreactor while using 99% less water (and possibly space).
- Bisabolol is one other useful chemical found naturally in the candeia tree, a plant native to the Amazon that has been reduced to endangered status. Now bioidentical bisabolol will be produced in any quantity using a bioreactor and the company’s proprietary enzymatic pathway.
- Plant and fruit fibers comparable to bananas will be converted into a substance called ‘soluble sweet fiber’, providing an alternative to other sugars and sweeteners; Allozymes received a $1 million grant to speed up this difficult process. Salehian reports yes with the results I made cookies and bubble tea.
I asked about the possibility of using microplastic-degrading enzymes, which have been the subject of many studies and which also appear in Allozymes promotional materials. Salehian said that while it’s possible, it is not currently economically feasible under their current business model – mainly, the customer would have to come to the company and say, “I want to pay to develop this.” But they have it on their radar and may soon get into recycling and processing plastics.
So far, all of this is more or less inside the company’s original business model of enzyme optimization as a service. However, the roadmap involves expanding the work from scratch, comparable to finding a molecule that matches the need, slightly than improving an existing process.
The enzyme customization service Allozymes provides is expected to be called SingZyme (as in single enzyme) and will proceed to be the entry-level option, filling the “we want to do it 100 times faster or cheaper” use case. A more expansive service called MultiZyme will take a higher-level approach, discovering or refining multiple enzymes to meet the more general premise of “we need something that does this.”
The billions of data points they collect through these services, nevertheless, will remain their mental property and will constitute “the largest library of enzyme data in the world,” Salehian said.
CEO Peyman Salehian and CTO Akbar Vahidi, co-founders of Allozymes. Image credits: Allozymes
“You can tell AlphaFold the structure and it will tell you how it folds, but it can’t tell you what will happen if it bonds with another chemical,” Salehian said, and in fact this response is the only element the industry is interested in. “There is no machine learning model in the world that can tell you exactly what to do because the data we have is so little and so fragmented; we are talking about 300 samples a day for 20 years” – a number that Allozymes machines can easily exceed in a single day.
Salehian said they are actively developing a machine learning model based on the data they have and have even tested it with a known result.
“We fed the data to a machine learning model and it got a suggestion for a new molecule that we are already testing,” he said, which provides a promising initial validation of the approach.
The idea is not unprecedented: we have covered many corporations and research projects that have shown that machine learning models will be very helpful in sorting through huge datasets, providing additional confidence even if their results cannot replace the real process.
The $15 million A round includes latest investors Seventure Partners, NUS Technology Holdings, Thia Ventures and ID Capital, in addition to repeat investments from Xora Innovation, SOSV, Entrepreneur First and Transpose Platform.
Salehian said the company is in excellent shape and has loads of time and money to realize its ambitions – except it might raise a smaller amount later this 12 months to fund its expansion into pharmaceutical products and open a U.S. office.