A team of researchers from the University of Natural Resources and Life Sciences in Vienna has developed artificial intelligence (AI) software that can describe the three-dimensional (3D) structure of therapeutic proteins. The aim of this research is to help companies in designing processes for purifying advanced therapies during manufacturing.
The software, currently being trained in geometry, will analyze the numerical values of various physical parameters of the proteins and link them to their molecular structure. By doing so, the AI can predict how the proteins will behave when purified during manufacturing processes.
Professor Johannes Buyel, the head of the research team, explained that the AI’s training is similar to teaching a computer to understand different types of doorknobs. Depending on the training dataset used, the AI might not be able to identify all types of doorknobs accurately. Therefore, the researchers are focusing on identifying the relevant physical parameters that are representative of the protein’s structure.
This AI development is an extension of previous research, but with the integration of the latest advancements in machine learning. In the past, researchers used molecular descriptors to correlate with experimental data, but it was a merely descriptive approach. Now, the team has developed a mechanistic model underlying the predictions.
In addition to assisting in process development for manufacturers, the AI software may enhance the understanding of protein biology. The ability to accurately describe the 3D structure of therapeutic proteins can have significant implications for the development and optimization of cell and gene therapies.
As this research progresses, manufacturers might be able to rely on AI predictions to improve their process development. This advancement could potentially revolutionize the manufacturing of advanced therapies, leading to more efficient and effective purification processes.
The source of the article is from the blog dk1250.com