Democratizing the science of protein design with Adaptyv Bio


Artificial intelligence-driven protein design promises to transform fields ranging from drug discovery to industrial materials development. But design only gets you so far – every invention needs to be tested in the real world as well. Enter the Swiss start-up Adaptive Biothat hopes its technology will solve this potential block in the innovation pipeline.

“Proteins are at the heart of the biorevolution, whether as new drugs, better enzymes for research and industrial applications, or as materials with new properties,” explains Adaptyv Bio’s CEO and co-founder Julian Englert. And the advance of AI solutions like DeepMind’s AlphaFold is breaking new ground in this regard, allowing researchers to automate the design of new proteins.

The problem, Englert explains, is that any new design then has to be tried out. “We’re getting very used to AI tools that generate content like text, but with that kind of output you can immediately tell if it’s working or if it’s rubbish,” he explains. “With an AI-designed protein, you can’t just look at it to see if it will work the way you want it to; it has to be tested in a laboratory.”

That’s where Adaptyv Bio comes in. The lab – Englert describes the facility as a “full-stack protein engineering foundry” – enables designers to quickly test new proteins on a large scale. They can send hundreds or even thousands of designs to Adaptyv Bio, which uses technologies such as robotics, microfluidics and synthetic biology to test them in the lab to determine which one works best. The process can be repeated over and over again to iterate the design until it is exactly what is needed. Each new test generates data that can be used to refine designs – and improve the subsequent performance of AI-powered protein design.

Englert believes the foundry could be a breakthrough for scientists by democratizing the protein technology industry. Currently, scientists without access to their own laboratories – largely reserved for the largest pharmaceutical companies and industrial conglomerates – wait months for the test results of each new design, at a significant cost.

Englert compares his company’s value proposition to the software development industry. “Imagine every time you used Github Copilot to generate code, you had to wait 10 weeks for it to run or to tell you it had a bug,” he says. “And imagine if each execution costs $1,000. That’s pretty much the situation for protein designers today.”

Adaptyv Bio was launched about two years ago and since then the company has been fine-tuning the technology at its foundry in Lausanne by collaborating with a number of customers in the biotechnology sector. “We’re now at a point where the technology works really well on an automated basis and we’re really ready to talk about it,” Englert added.

The company sees the drug discovery industry as its main target market, at least initially. In particular, small biotechnology companies, including “virtual biotech companies” that have raised funds for research but have no facilities of their own, will be able to use the Adaptyv Bio foundry to test the real-world efficacy of their protein designs.

However, protein design has applications in multiple fields, with scientists constantly working to develop new materials with specific properties. Industrial applications in the environmental sector are just one of the possible areas of interest.

“Look at the proteins that make up the incredibly powerful molecular machinery in every cell of our body,” says Englert. “Now imagine what kind of technological progress humanity could make if we could design new proteins for personalized medicines, industrial applications like new enzymes, or better, more sustainable materials.”

Englert and his co-founders met at the Swiss Federal Institute of Technology in Lausanne, where they founded Adaptyv Bio and developed the company as it progressed through the Y Combinator accelerator program. It then raised $2.5 million in funding from Wingman Ventures.