Pharma 5.0

Asymchem develops AI platform to optimise protein design in drug development

Published: 7-Nov-2024

Sequence Recommendation via Artificial Intelligence (STAR) addresses key challenges in protein design through intelligent algorithms and data-driven insights

CDMO Asymchem has created STAR, an artificial intelligence (AI) platform which aims to optimise the protein design process.

STAR was developed by Asymchem's Centre of Synthetic Biology Technology and AI team, and was created to improve the efficiency of protein engineering. 

The team responsible for STAR has since written a paper on the technology, which is available online (see reference list below).

According to Asymchem, a major challenge of protein design and engineering is the number of mutations that must be generated and screened. 

This process can take between three and six months, so there is an urgent need to reduce how long this process will take. 

By utilising machine learning, the platform can identify critical regions for mutation, meaning the number of experimental samples is reduced to a few hundred per round.

This means that the process can be complete in as little as one month, with researchers potentially discovering mutants with more than 50x better activity.

The STAR platform can offer:

  • Intelligent algorithms: By integrating active learning and virtual directed evolution, the system allows researchers to identify mutation sites beyond experimental data and recommends advantageous mutations
  • Data-driven insights: STAR has trained models with accurate predictive capabilities through years of proprietary protein engineering data
  • High-level automation: From data processing to model training and sequence recommendation, STAR offers a fully integrated, end-to-end protein design workflow

Asymchem’s CSBT team successfully applied the STAR platform to the evolution of glucose dehydrogenase, increasing the enzyme’s activity twofold and improving its stability tenfold.

They further used STAR to predict combination mutations, resulting in a fivefold increase in activity and another tenfold boost in stability.

Compared to traditional methods, STAR significantly reduced the time needed for enzyme evolution while improving efficiency in analyzing mutation relationships.
As the STAR platform continues to evolve, it will play a pivotal role in advancing Asymchem’s biopharmaceutical research, unlocking opportunities for breakthroughs in the field. 

 

Reference 

1  https://pubs.acs.org/doi/10.1021/acsomega.3c04832

Relevant companies

You may also like