Novasign Studio connects, models, optimizes and scales your whole bioprocess
One platform for the whole modelling chain. Bring equipment and file data together, design your experiments, build a hybrid model of each unit operation, test runs before you book them and carry the same model into scale-up.
Built for upstream, downstream, scale-up and process optimization across the full development lifecycle.
What is Novasign Studio?
Novasign Studio is one environment for bioprocess modelling, simulation and decision support. It connects experimental design, data preparation, model building and simulation, and your team works in it through the browser.
Plan your experiments
Define the process objective and design the experiments around the decision you need to make. Choose from nine types of design of experiments, including intensified DoE, where conditions change within a run as well as between runs, so only the runs that teach the model something new get reactor time.
Upload datasets or connect your sources
Upload CSV, Excel, Parquet or JSON files, or read process data straight from InfluxDB and Qubicon. Any other database is reached through a Python code node. Studio then aligns online and offline measurements on one time base. The preparation is a workflow, so it runs the same way for the next campaign.
Set up a model for your process unit
Combine mechanistic understanding with data-driven methods in a hybrid model of the bioreactor, tangential flow filtration or chromatography. Every training run is recorded with its parameters, metrics and lineage, so you can always trace where a model came from.
A hybrid model combines a given mechanistic structure (mass balances, stoichiometry) with a neural network that learns only the reaction kinetics.
Simulate before you book a reactor
Run a trained model across a parameter range, then compare and rank the scenarios in one view to see for which conditions it is worth performing the experiment.
One simulated run over 160 hours, the digital twin’s prediction as smooth lines and measured samples as dots that lie close to them. The product titer rises and levels off, while the impurity keeps rising toward the end of the run.
- Product titer, digital twin
- Product titer, measured
- Impurity, digital twin
- Impurity, measured
In Novasign, a bioprocess digital twin is a trained, validated hybrid model of a process or process segment that is run in-silico to predict process behaviour before running it physically. In a simulation, the feeds follow those of the runs the model was trained on.
Extend it with your own Python code
Where the mechanism is yours, write it in a Custom Python Code node of the same workflow, in a full code editor. Your code runs next to the Novasign models and its results flow into the next step like any other.
Why teams like working in Novasign Studio
A collaborative environment
Colleagues from different departments work on the same workflow at the same time. You see who is in the project and where they are working, and a node someone is editing is locked for everyone else until they are done.
- Two other participants are on the same canvas, each drawn as a colored pointer with their name beside it.
- Two nodes each carry a solid halo in their holder’s colour, with that person’s initials on the corner and a caption naming them.
- While a lock is held, nobody else can open that node’s dialog and nobody else can delete it.
- While your own pointer is over the drawing it is shown the same way the others are, labelled as you, and it returns to the ordinary cursor when you move away.
Direct access for your collaborators
Your collaborators open the project in their browser with nothing to install. A process engineer, a data scientist and a manager look at the same data, model and result instead of at copies sent around by email.
Traceability you can hand on
Every change carries its author and time, and every training run is recorded with its data, parameters and metrics. A result reopens as it was produced, so the whole team can follow a model and defend it.
Bring the process data together
Bring your data as CSV, Parquet, Excel or JSON files, exported from your instrument software, or connect a Qubicon system directly. Studio turns them into versioned, unit-aware datasets your team can query, profile and reuse. Each dataset stays available after training, so you can always show which data a model was built on. Take results out as files, or through Studio’s API.
No code where it helps, code where it matters
Process engineers build a workflow by dragging nodes and configuring each one in a dialog, and Studio checks every connection as it is made. Data scientists write their own Python in the same workflow, so both work on one platform. The AI assistant can help either of them on request.
Flexible deployment, on premises or in the cloud
Install Studio on premises, in your own cloud or on a dedicated appliance. On-premises and appliance deployments store process, model and dataset data inside your network, and an air-gapped installation sends nothing outside it and runs without any public internet exposure. Sign-in goes through your own identity provider.
- On-premise
- Your own data centreavailable today
- Your cloud
- Your own cloud accountavailable today
- Appliance
- A dedicated appliance you ownavailable today
- Managed
- Novasign Cloud, run by Novasignrolling out