← Back to Home

Roche Partners with Chai Discovery to Bring AI Protein Design into Multiple Drug Discovery Programs

Following several other major pharmaceutical companies, Roche will also adopt Chai’s models and platform. The collaboration expands the use of AI molecular design, but targets, deal terms and experimental results remain undisclosed, leaving key evidence still needed to demonstrate improvements in drug R&D efficiency.

By SURL BioNews

Whether AI can deliver practical value for drug R&D is increasingly being tested in pharmaceutical companies’ research programs. Chai Discovery announced a collaboration with Roche on November 5, adding another major pharmaceutical company setting for the AI company focused on designing protein therapeutics. The new development is that Roche will use Chai’s models and platform across multiple drug discovery programs; no results from these programs are currently available for assessment.

According to Chai’s announcement and the Business Wire press release republished by BioSpace, the collaboration spans different therapeutic areas at Roche, but does not specify diseases, molecular targets or the exact number of programs. Upfront payments, milestone payments and other financial terms were also not disclosed. Boris L. Zaïtra, Roche’s head of corporate business development, said Chai’s platform aligns with the company’s research strategy. The announcement did not provide further details on how the two companies will divide responsibilities or which stages of R&D will use the models.

Chai positions its technology as a computer-aided design tool for protein therapeutics: its models predict biological structures and functions to generate new molecular designs, helping researchers find candidate proteins that may bind to disease-related targets. These applications sit at the early end of drug discovery. The key is to turn computer-generated designs into molecules that can be tested experimentally, then progressively establish their biological activity and potential for development into drugs.

The announcement mentioned the latest model, Chai-3, and claimed strong target success rates and binding affinity, including against more challenging targets. However, these are the company’s descriptions of its platform’s capabilities. The collaboration announcement did not include Roche validation data, the set of targets tested, control methods or information about training data. It therefore provides no basis for assessing the model’s performance in Roche’s programs or estimating how much time and cost it could save.

Even if candidate molecules bind tightly to a target, researchers must still establish whether that binding produces the intended effect and whether the molecules can be manufactured consistently, maintain appropriate exposure in the body and have acceptable safety. The announcement disclosed no drug candidates or human study data, so the collaboration cannot yet be viewed as evidence of efficacy. Nor can it establish whether the quality, nonclinical and clinical evidence needed for future regulatory submissions is already in place.

Background

FirstWord HealthTech noted that before Roche joined, Chai had already established collaborations this year with Eli Lilly, Pfizer, Novartis, Bristol Myers Squibb and GSK. Compared with the $400 million financing announced in July, this news represents a different form of progress: continued adoption of the platform by pharmaceutical companies. However, growth in the list of collaborators and improvements in R&D output remain two matters that need to be measured separately.

For Roche, the agreement also sits within a broader AI R&D strategy. FirstWord reported that the company is planning autonomous AI laboratories, but the available information does not explain whether Chai’s platform will connect with these facilities. The next more informative milestone will be whether the two companies can disclose specific targets, reproducible experimental results and progress in advancing designed molecules into further development, allowing platform adoption to gradually build into evidence for drug R&D.

References

  1. Contract Pharma
  2. Chai Discovery
  3. FirstWord HealthTech
  4. BioSpace (Business Wire press release)