Immunity-boosting monoclonal antibodies are a wonder of modern medicine, lab-made proteins that can treat cancer, autoimmune diseases, and many other conditions. The market for these therapies is projected to be Double by 2030it might seem that the only thing they can’t do is grow fast enough.
A new study from the University of Oklahoma also aims to end that limitation.
in study Published in a magazine communication engineering, Chongul breadOU Professor of Computer Science and Biomedical Engineering, and Wang PenghuaA doctoral student in data science and analytics, he details a machine learning model that dramatically accelerates monoclonal antibody manufacturing schedules.
“We are solving a critical bottleneck in the biomanufacturing production process,” Wang said. “The key is to get to market faster.”
In humans, antibodies are produced by white blood cells known as B cells, but in biomanufacturing, that role falls to Chinese hamster ovary cells (CHO), the industry standard for producing therapeutic antibodies.
Pan said the process isn’t that different from making beer. In fermentation, yeast feeds on sugar, which is converted into alcohol. Similarly, harvested CHO cells receive nutrients designed to maximize antibody production.
However, not all clonal cell lines produce antibodies at the same rate. Productivity varies. To select the highest yielding cell lines, biomanufacturers must screen culture samples, and the production stage can take several weeks. Accelerating this timeline is a priority for pharmaceutical companies and could help reduce drug costs for patients.
Pan and Wang hypothesized that proliferation data obtained during the early stages of production could be used to predict cell productivity. To test their theory, they partnered with an Oklahoma City-based company. Wheeler’s careera contract development and manufacturing organization (CDMO) focused on antibody therapeutics. Wheeler provided production data combined with established mathematical equations (the Luedeking-Piret model, which describes how cells grow and produce proteins) to train and validate the machine learning tools.
After testing and fine-tuning, the researchers’ model accurately selected the better-performing clone in 76.2% of trials and accurately predicted daily production trajectories from days 10 to 16 using only data collected during the first nine days of growth. The results demonstrate the effectiveness of “simulated real-world clone selection,” which ultimately allows for “faster and more reliable identification of high-performing clones,” Pan and Wang wrote.
Pan said further testing and training of the model is needed before it can be introduced into Wheeler’s production process. But company officials said they were encouraged by early results.
“Wheeler Bio is committed to leveraging artificial intelligence and machine learning to accelerate approaches to cell line development and process development for manufacturing antibody therapeutics,” said Patrick Lucey, president and CEO of Wheeler Bio.
“This foundational work is the first step in Wheeler’s strategic effort to leverage artificial intelligence and machine learning to further enhance our capabilities. ModularCMC™ Platform”
This study $35 million program Funded by the U.S. Economic Development Administration to expand the Oklahoma City region’s biotechnology industry cluster. OU’s Gallogly College of Engineering and the recently opened OU Bioprocessing Core Facility is the lead institution in this effort, which aims to link academic innovation with industrial applications.
“In academia, there is a tendency to pursue theoretical research,” Pan says. “However, this research and our partnership with Wheeler Bio gives us the opportunity to apply our machine learning and data science expertise to real-world problems facing the industry.”
reference: Wang P, Verma D, Chiu Y et al. Luedeking-Piret regression for multi-step ahead prediction and clone selection in monoclonal antibody biomanufacturing. communication engineering. 2025;4(1):220. doi: 10.1038/s44172-025-00547-7
This article has been reprinted from the following material: Note: Materials may be edited for length and content. Please contact the citation source for details. You can access our press release publishing policy here.
