Swiss Adaptyv Biosystems Sàrl has launched two fast-feedback tools that allow cheap, rapid and minaturised validation of AI-designed proteins.
Machine learning algorithms and generative AI are upending traditional processes in life sciences and collapsing time frames in drug discovery and materials development. Lausanne-based Adaptyv Biosystems’ full stack protein engineering foundry is aimed at paving the way for protein designers to develop new medicines, novel enzymes and sustainable materials. According to the company, its tools are an answer to costly and timely services that find bugs in AI-generated protein structure designs based on novel algorithms such as Alphafold or RFDiffusion.
AlphaFold by DeepMind predicts the 3D structure of a protein from its amino acid sequence and has been used by more than 1 million researchers over the 18 months in which it has been publicly available. Since then a plethora of other AI tools have emerged, such as the recently open-sourced RFDiffusion, a machine learning model which allows researchers to generate computational protein designs using just their laptop.
As AI thus makes rapid progress in the world of bits, translating those computational designs into physical, functioning proteins remains challenging. As a next step in the evolution of AI-generated protein structures, Adaptyv Bio is building a full-stack platform to allow protein engineers to validate their AI-generated protein designs.
According to Julian Englert, CEO and co-founder at Adaptyv Bio, which baged CHF2.5m in pre-seed financing led by Wingman Venture, “proteins are at the core of the biorevolution, whether in the form of new medicines, better enzymes for research and industrial applications or as materials with novel properties. Validating your protein designs in the lab to see if they work remains a huge pain. Imagine every time you used Github Copilot to generate some code you had to wait 10 weeks for it to execute or to tell you that it had a bug. And imagine each execution costs 1000 USD. That’s pretty much the situation for protein designers today.”
By making it easier to generate data about how well the designed proteins work, Adaptyv Bio allows protein engineers and AI models to get more feedback about their designs and helps them steer towards better performing proteins. At the heart of Adaptyv Bio‘s foundry are protein engineering workcells — custom, automated setups that miniaturise processes usually requiring multiple laboratory machines, instead conducting them in parallel across tiny microfluidic chips that consume 1,000 times fewer reagents than any commercially available alternative. Users can write experimental protocols or have AI write them for them and the workcells then execute the experiments fully autonomously while tightly controlling and tracking the experiments’ parameters. All measurement data is automatically processed and uploaded to allow users to improve their machine learning models with each experiment.
"Over the next 12 months, we plan to scale up our lab further and increase the number of protein design applications we can support," says Englert. "We also just opened up early access for users to submit their protein design projects to us and we’re trying to onboard new projects as soon as possible”
In order to accelerate the field of protein engineering as a whole, Adaptyv Bio hasoopen-sourced two of the company’s internal tools: ProteinFlowis a Python library to allow protein designers to easily create high-quality datasets for better AI models, and Automanceris, an extensible software platform to run automated experiments, enabling researchers to build their own experimental protocols and integrate different laboratory instruments.