Adaptyv Bio launched to ‘transform protein engineering’

AI tools like ChatGPT are revolutionizing the way the world generates text, images and code. In the same way, machine learning algorithms and generative AI reverse traditional processes in the life sciences and tear down timeframes in drug discovery and materials development.

AlphaFold by DeepMind is probably the best-known machine learning model in this space. It predicts the 3D structure of a protein from its amino acid sequence and has been used by more than 1 million researchers in 18 months. Since then, other AI tools have emerged, such as the recently open source RFDiffusion, a machine learning model that allows researchers to generate computational protein designs using a laptop.

However, translating such computational designs into functioning physical proteins remains challenging. Adaptyv Bio has been launched to address this with its protein casting. Combining robotics, microfluidics and synthetic biology techniques, Adaptyv Bio is building a full-stack platform to enable protein engineers to validate their AI-generated protein designs.

Validating protein designs

Julian Englert, CEO and co-founder at Adaptyv Bio, said: “Proteins are at the heart of the biorevolution, whether in the form of new drugs, better enzymes for research and industrial applications, or as ingredients with new properties. As a protein designer, you now have access to great new AI tools like AlphaFold or RFDiffusion. But validating your protein designs in the lab to see if they work is still a huge pain.

What AI models need most is data – to train them and to improve the predictions they make. By making it easier to generate data about how well designed proteins work, Adaptyv Bio enables protein engineers and AI models to get more feedback on their designs and helps them lead to better performing proteins.

“Think of AI in a self-driving car. To keep the car on the road and get it to its destination, the AI ​​model needs to have tight feedback by getting lots of high-quality data from the car’s camera sensors. The principle is much the same for AI models designing new proteins, only the feedback mechanism here is actually creating the protein in our lab and testing its performance,” added Englert.

Adaptyv Bio cells work

At the heart of Adaptyv Bio’s foundry is the protein engineering work cell – an automated, customized setup that miniaturizes processes that would normally require several laboratory machines, instead of doing them in parallel across tiny microfluidic chips. The user can write an experimental protocol (or have an AI write one for it) and the workcell then runs the experiment completely autonomously while controlling and tracking the experiment parameters. All measurement data is automatically processed and uploaded to allow users to refine their machine learning models with each experiment.

Says Englert: “Our work cells are fully automated, consume 1,000 times less reagents than commercially available alternatives, and we can run thousands of different proteins per day on each individual setup. To streamline experimental workflow, we have developed many specialized synthetic biology and automation techniques. Over the next 12 months, we plan to further upgrade our lab and increase the number of protein design applications we can support. We also recently opened up early access for users to submit their protein design projects to us, and we are trying to get new projects in as soon as possible.”

To accelerate the protein engineering field as a whole, Adaptyv Bio is also open-sourcing two of their in-house tools, which are already starting to catch the attention of researchers and engineers in the field.

ProteinFlow is a Python library to allow protein designers to easily create high-quality datasets for better AI models. Automancer is an extensible software platform for running automated experiments, enabling researchers to build their own experimental protocols and integrate different laboratory instruments.

“Our mission is to make protein engineering easier and enable more researchers to design new proteins. Take a look at the proteins that make up the powerful molecular machinery inside every cell in our bodies. Now imagine the kinds of technological advancements humanity could make if we could start designing new proteins for personalized pharmaceuticals, industrial applications like new enzymes, or better and more sustainable materials,” concludes Englert.

Adaptyv Bio was founded by a group of engineers from EPFL, the Swiss Federal Institute of Technology in Lausanne. In 2022, they raised $2.5 million in an seed funding round led by Wingman Venture, after going through startup accelerator Y Combinator.

The company is based at the newly built Biopole life sciences campus in Lausanne, Switzerland.