Lab-in-the-loop
We validate Trellis process recommendation in our own wet lab, using real-world feedback to fine-tune the tool.
Wander into unmapped design spaces, find the shortest path to a robust bioprocess, and confidently go from lab to scale.
Bioprocessing to a living, breathing system. A system that is affected by the multidisciplinary (biology, chemistry, mathematics and physics) factors while still being guided by economics.
While other disciplines received the gift of computation, a scientist is still disadvantaged by trial and error, experience, and heaps of data.
From recommending the design of experiment (DOE), building protocols, ingesting the experimental data, to recommending the next design space, Trellis’ suite of bioprocessing toolbox helps you get to controllable bioprocess, and scale-up faster and with fewer runs.
OFAT, DoE and Trellis were benchmarked on the same virtual 12-factor CHO fed-batch process. Trellis reached effectively the full known optimum; DoE plateaued near 77%, OFAT near 50%.
The difference is where the experiments go. OFAT searches locally, DoE spreads broadly and wastes runs far from the optimum, Trellis explores uncertain regions and concentrates where better performance is most likely.
% of known optimum reached on a simulated 12-factor CHO fed-batch process.
Trellis combines scientific reasoning with machine learning. AI orchestrates a set of agents that build your models, and those models account for bioprocess mechanics that most dry-lab tools leave out.
Trellis handles design spaces with up to 50 parameters and explores the entire space, with every parameter in play at the same time, until it gets you to the true global optima.
Tell Trellis your target response, your run budget and your setup. It selects the right sampling methodology and objective, then recommends the right DoE across the entire design space, so you run the least number of experiments.