A paper in Biotechnology and Bioengineering, first published online on 13 October 2026 by authors affiliated with GSK, describes a digital twin that decides when to transfer cells from the N-1 seed bioreactor to the production bioreactor of a commercial monoclonal antibody process. The full text is behind a subscription, so this brief is limited to what the abstract says.
The problem: seed density varies
The N-1 seed bioreactor is the stage where cells grow before the transfer to a production bioreactor. According to the abstract, the final viable cell density (VCD, the number of living cells per unit volume) at this stage often varies. That leads to variable seeding densities in the production bioreactor and to variable total protein output per batch.
What the authors built
The authors developed an upstream, stage-specific, level-2 digital twin, which they also call a digital shadow. A cell growth process model predicts growth and updates its forecasts using off-line and in-line measurements. This lets the team determine the final sample time precisely.
What the abstract reports
- After deployment, all N-1 to production transfers fell within acceptable limits.
- Harvest yield showed statistically significant increases relative to historical batches.
- The authors attribute the gain to significantly higher on-target accuracy of the final N-1 VCD and significantly higher working volumes in the production bioreactors.
The abstract gives no numerical values for the limits, the yield increase or the accuracy, so none are reported here. It also does not name the product, the scale or the sites.
What it means for a plant
The abstract frames the twin as a way to make the seed transfer land on target, which carries through to production bioreactor working volume and harvest yield. It says the approach lays a foundation for more advanced process control strategies. Before copying it, a plant would need the full paper for the model structure, the specific off-line and in-line measurements used and the size of the yield effect, none of which the abstract states.
Sources
- Biotechnology and Bioengineering, A Biomanufacturing Digital Twin for Consistent N-1 Seed Bioreactor Transfers for Commercial Monoclonal Antibody Production, 13 Oct 2026 (abstract read via Europe PMC).
