- Asymchem has launched an AI-driven platform named STAR, aimed at addressing challenges in protein design.
- The STAR platform significantly reduces the time required for protein engineering, delivering faster results with higher efficiency.
Asymchem, a contract development and manufacturing organisation (CDMO), has introduced an innovative artificial intelligence (AI) platform to transform protein design. Known as STAR (Sequence Recommendation via Artificial Intelligence), this platform aims to tackle the time-consuming challenges associated with protein engineering. Developed by Asymchem’s Center of Synthetic Biology Technology (CSBT) and AI team, the platform was designed to streamline the biosynthesis research and development process for biopharmaceuticals.
STAR addresses one of the major hurdles in protein engineering—the lengthy process of generating and screening mutations. Traditionally, this process can take between three to six months. However, the STAR platform accelerates the process dramatically by using machine learning to pinpoint critical mutation sites, reducing the number of experimental samples required and speeding up the discovery of improved proteins. In some cases, researchers have seen a 50-fold improvement in activity within just one month.
The STAR platform offers several key features. It integrates advanced machine learning algorithms, including active learning and virtual directed evolution, allowing researchers to identify mutation sites and recommend beneficial mutations beyond what experimental data alone can provide. With years of proprietary data, the platform can offer highly accurate predictive capabilities to guide protein engineering efforts. Furthermore, STAR’s fully automated end-to-end workflow—ranging from data processing to model training and sequence recommendation—enables seamless execution of protein design projects.
In practical application, the STAR platform has already delivered remarkable results. Asymchem’s CSBT team successfully used it to evolve glucose dehydrogenase, improving the enzyme’s activity twofold and its stability tenfold. Through predicting combination mutations, they achieved a fivefold increase in activity and an additional tenfold improvement in stability, all while cutting down the time required for these results.
Overview of outsourcing trends 
