Amgen dons BigHat to bring AI insights to next-gen antibodies

Amgen is advancing its collaboration with AI-enabled antibody design shop BigHat Biosciences. With the first phase of the previously undisclosed partnership validating the technology, Amgen has tasked BigHat with creating a lead panel of VHH antibodies.

BigHat put itself on the map early last year when it disclosed $19 million in series A funding from a VC syndicate led by ‍Andreessen Horowitz. The funding positioned BigHat to scale up a platform that pairs a wet lab with AI and machine learning to accelerate the validation of antibody design hypotheses. While the antibody design space is heavy with proven technologies, BigHat is betting that its approach will have an edge when applied to next-generation antibodies such as VHHs and bispecifics.

“Technologies that screen blood for antibodies work great for vanilla monoclonals. But with these more engineered next gens, you need technology that can learn from the data and introduce changes to an antibody, so that you can make high-quality, clinical-grade, next-generation antibodies with all the human-engineered features that make them next generation,” BigHat CEO Mark DePristo, Ph.D., said. 

The deal with Amgen, a pioneer of next-generation antibody modalities such as bispecifics, gave BigHat a chance to show its technology can live up to that billing. BigHat began working with Amgen around one year ago, having come into its orbit when it won an Amgen Golden Ticket in 2020, and has now finished the first phase of the project.

RELATED: BigHat bags $19M to 'supercharge' antibody development

In the first phase, BigHat worked to show its platform can improve the properties of a next-generation antibody, in this case a VHH camelid antibody fragment. The completion of that project has unlocked the second phase of the collaboration. 

“The second phase really opens up the platform to let it search, optimize as much as possible on many properties, not just biophysical properties like affinity, but also stability, aggregation,” DePristo said. “That's allowing us to produce a molecule that's ready for scale up and downstream testing and then heading off into the clinic.” 

Peyton Greenside, Ph.D., chief scientific officer at BigHat, said the platform enables the generation and testing of designs for hundreds of antibodies a week. The pace of the turnaround means BigHat can run through many more cycles of the design-build-test process than is possible using conventional technologies. As the cycle spins, BigHat seeks to overcome challenges encountered when working with complex designs. Each cycle generates data to inform both the antibody being tested and future projects, as determinants of properties such as aggregation and thermostability are generalizable across molecules. 

As well as working with Amgen and seeking to finalize deals in its business development pipeline, BigHat is using the platform to support internal programs. BigHat has a therapeutic team and multiple assets in development.