Digital Twin

Novasign's Digital Twin is a desktop application designed to accelerate bioprocess development through advanced in-silico simulation. By leveraging a hybrid model-based digital twin simulator, it enables scientists and bioprocess engineers to predict and explore a vast range of potential process scenarios within minutes, eliminating the need for costly and time-consuming physical experiments.

The platform is built for life sciences teams seeking to optimize biopharmaceutical manufacturing processes — from cell culture growth phases through to production — with the ability to incorporate economic parameters such as cost per dose and space/time yield alongside traditional metrics like titer and yield.

Advanced Customization Options

• X-Axis Customization: Users can rename and set durations for each stage on the X-axis, enabling precise time series analysis and the ability to adjust time scales to reflect distinct experimental stages such as growth and production phases.
• Constant Variables: Define constant parameters for each stage — for example, iDoE factors — allowing conditions to remain consistent within a stage while varying across stages (e.g., different temperature settings per stage), making it straightforward to explore diverse process scenarios.
• Simulated Variables: Introduces variables that change based on custom inputs, covering an unlimited range of process scenarios. Users can also incorporate economic parameters into models to optimize for space/time yield or cost per dose rather than titer alone.

Robust Simulation Framework

• Multi-Stage Processing: The process timeline can be segmented into distinct stages, each with specific conditions. This supports intensified designs of experiments by varying conditions across stages, enabling detailed analysis and optimization within a single simulation run.
• Multithreaded Simulation: The Digital Twin leverages multithreaded processing to simulate all possible combinations of constants across different stages simultaneously, significantly reducing computation times and enabling rapid forecasting of complex experimental designs.
• Portable Database for Resource Management: Simulations are stored in a lightweight, portable database requiring no installation, offloading data to disk to conserve RAM. Completed simulations are immediately saved and made accessible for analysis without extensive memory allocation, allowing numerous simulations to run without impacting system performance.
• Templating and Reusability: Digital structures can be reused across projects. Users can update hybrid models at any time, copy entire project configurations including custom code, and paste them into new projects ready to use out of the box.

Interactive Visualization and Exploration

• Interactive Sliders: The visualization section provides sliders for all desired variables, with each slider representing a different process stage. Users can dynamically adjust parameter values across stages and immediately see the impact on outcomes, with corresponding simulations fetched from the database and plotted in real-time.
• Variable Selection for Plotting: Users can select which variables to display, enabling or disabling them as needed to focus on specific aspects of the simulation.
• Series Type and Independent Y-Axes: Each variable can be plotted as a line series or scatter plot, and each can be assigned an independent Y-axis for detailed cross-parameter comparison.
• Axis Range Configuration: Axes can be fixed to specific minimum and maximum ranges per variable, facilitating consistent comparison across different simulation runs.
• Confidence Intervals: For simulations incorporating models from multiple modes, the plotting menu supports the display of confidence intervals for each variable, deepening analytical insight.

Process Optimization and Search Functionality

• Variable Selection for Optimization: Users can specify any parameter of interest — such as titer, yield, or cost-associated factors — for further analysis and optimization.
• Search Criteria: The software supports refined searches based on minimum and maximum values of selected variables. Users can specify whether to find these extremes at the end of the process, at the end of specific stages, along the entire time series, or up to a certain stage, providing flexible assessment of results.
• Result Listing: All lookup results are listed clearly, allowing users to quickly grasp the range of outcomes meeting their specified criteria.
• Tolerance Percentage: A tolerance percentage feature broadens the scope of optimization by identifying additional simulations that fall within a specified percentage of the best result, accommodating natural process variability.
• Easy Visualization of Results: Clicking the Plot button next to any listed result automatically loads the corresponding process into the plot area, with sliders adjusted to reflect the exact conditions of that simulation.

Novasign has announced a strategic partnership with Repligen to further advance bioprocessing digitalization, underscoring the platform's relevance within the broader bioprocess development ecosystem. The Digital Twin is positioned as a key tool for teams looking to reduce experimental costs, accelerate development timelines, and make data-driven decisions across the full bioprocess design space.