Optimizing Culture Media Using Small Datasets via New Computer Model

As featured in A*STAR newsroom and Genetic Engineering and Biotechnology News.

A new spinoff company is about to launch a computer model that aims to accelerate process development by speeding up the optimization of cell culture media. AuctuCel, which plans to open for business on June 1, will offer a mechanistic model that requires less client data to work than those based on artificial intelligence (AI), according to Zach Pang, PhD, group leader at the A*STAR Bioprocessing Technology Institute in Singapore.

Pang, who helped develop the model and introduced it at Bioprocessing Summit Europe in March, said, “Our key message is that you no longer have to perform laborious, time-consuming, manual optimization for culture media. Our new computational approach gives you an option that’s much faster and cheaper, helping accelerate your entire development timeline.”

Pang explained that the team’s deep knowledge of culture media helped them to develop a mechanistic model, which doesn’t require large training datasets to work.

“It doesn’t merely rely on correlation or a statistical method,” he added. “The person who runs our model is a hard-core biologist who’s taken up mechanistic modeling as, for any mechanistical model, you need to know what’s going on.”

The model can track what enzymatic reactions are present or absent, as well as predict intracellular fluxes of ingredients in the media. It uses this information and genetic details about the cell line to map out cellular metabolism.

At the Summit, Pang, who claimed the model has a predictive accuracy of more than 80%, presented a case study involving Chinese Hamster Ovary (CHO) cells, a type of cell commonly used in biopharmaceutical production. Explaining the benefits of the model compared to AI, Pang said, “Ours isn’t a ‘black box’ method. We know what’s happening metabolically and can have a scientist or bioengineer validate what we’re seeing.”

Although not designed to eliminate experiments, the model can reduce the number needed, noted Pang. “As a wet lab scientist, I can sympathize with both sides of this argument,” he said. “Experiments are expensive.”

By reducing the amount of data needed to understand cell culture media, he hopes the new model will help companies cut costs.

Source: Genetic Engineering and Biotechnology News - Optimizing Culture Media Using Small Datasets via New Computer Model