What Is FedBatchDesigner?
FedBatchDesigner is a free, browser-based dashboard for modeling and optimizing growth-arrested fed-batch (two-stage fed-batch, 2SFB) processes. Developed by Graf, Libiseller-Egger, Gotsmy, and Zanghellini at the University of Vienna and published in ACS Synthetic Biology in 2025, the tool lets you explore how different feeding strategies and growth-to-production switching times affect your process's titer, rate, and yield (TRY) without writing a single line of code.
The core idea behind a growth-arrested fed-batch process is separating biomass accumulation from product synthesis into two distinct phases. In the first stage, cells grow rapidly on a complete medium. A deliberate switch (nutrient limitation, temperature shift, or genetic trigger) then arrests growth, and the non-growing biomass channels carbon toward the target product. FedBatchDesigner automates the process design question: given my organism's kinetics, what feed rate and switching time maximize the metric I care about most?
Quick Verdict
Tool Snapshot
| Developer | University of Vienna (Zanghellini Lab) |
| Access | Free, no registration, browser-based |
| URL | chemnettools.anc.univie.ac.at/FedBatchDesigner |
| Source Code | GitHub (MIT license) |
| Stack | Shiny for Python v1.3.0 + Plotly v6.0.1 |
| Best For | Two-stage growth-arrested processes in microbial hosts |
| Not For | Continuous growth-production coupling, perfusion, mammalian cell culture |
| Paper | Graf et al. 2025, ACS Synth. Biol. 14(8):3252-3257 |
Inputs and Setup
FedBatchDesigner requires three categories of input, all entered through a sidebar panel. No data files, no programming environment, and no account creation needed.
The production kinetics follow the Luedeking-Piret model, which separates product formation into a growth-associated term (proportional to growth rate) and a non-growth-associated term (proportional to biomass). This is the standard framework for describing microbial product formation and maps directly onto the two-stage concept: the growth-associated term dominates during the growth phase, while the non-growth-associated term drives production during growth arrest.
Jupyter Notebooks for estimating these parameters from your own experimental data are provided in the GitHub repository.
The Three Feeding Strategies
FedBatchDesigner evaluates three production-phase feeding strategies and compares each against the full grid of feed rates and switching times.
| Strategy | Feed Rate Profile | Best When | TRY Bias |
|---|---|---|---|
| Constant | F = F0 (fixed) | Simple equipment, low-risk processes | Balanced titer and yield |
| Linear | F(t) = F0 + k·t | Increasing biomass with moderate metabolic load | Higher titer than constant |
| Exponential | F(t) = F0·eμset·t | Maximum substrate utilization, high qp organisms | Highest productivity, lower titer |
The analytical solution is the key differentiator from general-purpose bioprocess simulators. Because the growth-arrested assumption simplifies the differential equations (substrate concentration is approximately zero throughout both feed phases), FedBatchDesigner avoids numerical integration entirely. This makes the exhaustive TRY landscape evaluation near-instantaneous in the browser.
Outputs and Interactive Visualizations
FedBatchDesigner generates four interactive Plotly panels that together give you a complete picture of your process design space.
- Space-time yield heatmap (top-left): volumetric productivity as a function of feed rate and the fraction of total feed volume spent in the growth phase. The optimal point is marked with a black star.
- Final titer heatmap (bottom-left): product concentration at harvest, same axes. Shows the titer-productivity trade-off spatially.
- Productivity-titer trade-off curve (top-right): the Pareto frontier for each feeding strategy overlaid. This is where you see how much titer you sacrifice for each unit of productivity gain.
- Time-series trajectories (bottom-right): biomass, product concentration, and reactor volume over the process. Click any point on the heatmaps to see its corresponding time course.
Clicking any point on a heatmap (purple circle) compares it side-by-side with the computed optimum (black star). Summary tables show key metrics for both processes, and everything is exportable as CSV or PNG.
Walkthrough: L-Valine Production in E. coli
The l-valine case study from the original paper (based on experimental data from Hao et al. 2020) is the best way to understand FedBatchDesigner in practice. Here is the step-by-step workflow.
Worked Example: L-Valine in E. coli W3110
Organism: E. coli K-12 W3110, engineered for l-valine overproduction
Growth arrest trigger: Microaerobic shift (reduced oxygen supply)
Reference process: Hao et al. 2020 (standard fed-batch, 86 g/L titer, 0.33 g/g yield, 1.95 g/L/h productivity)
Step 1: Enter reactor parameters
- Load the E. coli default parameter set
- Set batch duration to 4 h and growth-phase μ to 0.52 h-1 (matching Hao et al.)
- Enter batch and feed substrate concentrations from your medium recipe
Step 2: Enter production kinetics
- Set the non-growth-associated qp for the microaerobic production phase
- Enter the product yield on substrate during growth arrest
- The Jupyter Notebook in the GitHub repo can estimate these from time-series data
Step 3: Evaluate the landscape
- The heatmaps populate immediately (analytical solution, no wait)
- The constant-feed optimum: 84 g/L titer, 0.41 g/g yield, 2.33 g/L/h productivity
- Switch to exponential feed: 78 g/L titer, 2.79 g/L/h productivity (+20%)
Step 4: Interpret the trade-off
Constant feed: Titer = 84 g/L | Rate = 2.33 g/L/h | Yield = 0.41 g/g
Exponential: Titer = 78 g/L | Rate = 2.79 g/L/h | Yield ~0.38 g/g
Reference: Titer = 86 g/L | Rate = 1.95 g/L/h | Yield = 0.33 g/g
The exponential strategy trades 7% titer for 20% higher productivity. The constant strategy matches the reference titer while improving yield by 24%. FedBatchDesigner's click-to-compare feature makes this trade-off immediately visible.
The tool also revealed that extending the growth phase from 12 h (as in the published protocol) to 17 h yields only a marginal 5% productivity improvement, suggesting that the original switching time was already near-optimal for this organism.
Case Study: Ethanol Under Nitrogen Starvation
The ethanol case study illustrates when growth-arrested fed-batch does not automatically win. Using S. cerevisiae with ATP wasting (Zahoor et al. 2020), FedBatchDesigner showed that a conventional one-stage fed-batch still achieves higher volumetric productivity than a two-stage process with the current production-phase kinetics.
The critical finding: the growth-decoupled ethanol formation rate would need to increase by at least 3-fold before a 2SFB strategy outperforms single-stage. This is exactly the kind of insight that saves months of wasted bioreactor time. If your organism's non-growth-associated qp is too low relative to the growth-associated term, growth arrest hurts more than it helps.
The ethanol case study was based on shake-flask data rather than bioreactor experiments, so FedBatchDesigner served as a pre-experimental planning tool to inform what the first 2SFB bioreactor runs should look like and whether they were worth attempting at all.
How FedBatchDesigner Compares to Other Tools
FedBatchDesigner occupies a specific niche in the bioprocess design tool landscape. It is the simplest tool to use but the most constrained in scope.
| Feature | FedBatchDesigner | OptFed | OptMSP |
|---|---|---|---|
| Scope | 2-stage growth-arrested only | General fed-batch optimization | Arbitrary multi-stage bioprocesses |
| Interface | Browser (no code) | Python/MATLAB | MATLAB toolbox |
| Solution method | Analytical (exact) | Nonlinear dynamic optimization | Sequential dynamic optimization |
| Speed | Near-instantaneous | Seconds to minutes | Minutes |
| Cost | Free | Free (open-source) | MATLAB license required |
| Organism flexibility | Any (custom parameters) | Any (custom ODE models) | Any (custom ODE models) |
| Growth-production coupling | Decoupled only | Coupled or decoupled | Coupled or decoupled |
| Interactive visualization | Yes (Plotly heatmaps) | Limited | Limited |
Limitations and When Not to Use It
FedBatchDesigner is deliberately narrow in scope. Understanding its boundaries prevents misapplication.
- Growth-arrested only. The model assumes zero growth during the production phase. If your product is partially growth-associated and you cannot fully arrest growth (common in mammalian cell culture), the tool's predictions will be inaccurate.
- No kinetic parameter estimation built in. You need pre-existing values for μmax, Yxs, maintenance coefficient, and Luedeking-Piret parameters. The supplementary Jupyter Notebooks help, but require basic Python familiarity.
- Substrate-limited only. The model assumes the limiting substrate concentration stays near zero throughout both feed phases. If your process accumulates substrate (e.g., high-glucose batch phases), the analytical solution breaks down.
- No product inhibition or toxicity. The model does not account for product feedback inhibition. For products like ethanol, acetate, or organic acids where high titers inhibit the producing cells, actual performance will diverge from predictions at high concentrations.
- Single product. Co-product formation (e.g., acetate overflow during l-valine production) is not modeled explicitly, though its effect can be partially captured through the yield parameters.
Fed-Batch Calculator
Design your fed-batch feeding profile, calculate feed rates, and predict substrate trajectories for both growth-coupled and growth-arrested processes.
E. coli Expression Optimizer
Optimize IPTG concentration, induction temperature, and OD for maximum soluble protein expression in E. coli.
Related Tools
- Fed-Batch Calculator - Calculate exponential and constant feed rates for E. coli, CHO, and yeast fed-batch processes
- E. coli Expression Optimizer - Optimize induction conditions for maximum soluble protein yield
- OTR/kLa Estimator - Estimate oxygen transfer rates for aerobic-to-microaerobic transition planning
Frequently Asked Questions
What is FedBatchDesigner and what does it do?
FedBatchDesigner is a free, browser-based tool from the University of Vienna that optimizes growth-arrested (two-stage) fed-batch processes. You input basic reactor and organism parameters, and it exhaustively evaluates constant, linear, and exponential feeding strategies across a grid of feed rates and switching times to identify the optimal titer-rate-yield trade-off.
What inputs does FedBatchDesigner require?
FedBatchDesigner requires reactor parameters (initial and maximum volume, batch and feed substrate concentrations), organism physiology (maximum specific growth rate, biomass yield on substrate, maintenance coefficient), and production kinetics (growth-associated and non-growth-associated specific product formation rates). Default parameter sets are available for E. coli K-12 W3110 and S. cerevisiae.
Is FedBatchDesigner free to use?
Yes. FedBatchDesigner is completely free, requires no registration, and runs entirely in the browser. The source code is MIT-licensed and available on GitHub, so you can also run it locally or modify it for your specific needs.
Can FedBatchDesigner handle any organism or just E. coli?
FedBatchDesigner works with any organism where growth can be arrested and production decoupled. It ships with default parameters for E. coli and S. cerevisiae, but you can input custom kinetic parameters for any host. The tool has been demonstrated for l-valine production in E. coli (microaerobic switch) and ethanol in S. cerevisiae (nitrogen starvation).
How does FedBatchDesigner compare to OptFed or OptMSP?
FedBatchDesigner focuses specifically on growth-arrested two-stage processes with analytical solutions for speed, while OptFed uses nonlinear dynamic optimization for general fed-batch problems and OptMSP handles arbitrary multi-stage bioprocesses. FedBatchDesigner is the simplest to use (browser-based, no coding) but limited to two-stage growth-arrested scenarios. Choose OptFed or OptMSP when you need continuous growth-production coupling or more than two stages.
References
- Graf AC, Libiseller-Egger J, Gotsmy M, Zanghellini J. Fedbatchdesigner: A User-Friendly Dashboard for Modeling and Optimizing Growth-Arrested Fed-Batch Processes. ACS Synth. Biol. 2025;14(8):3252-3257. doi:10.1021/acssynbio.5c00357
- Hao Y, Ma Q, Liu X, Fan X, Men J, Wu H, Jiang S, Tian D, Xiong B, Xie X. High-yield production of L-valine in engineered Escherichia coli by a novel two-stage fermentation. Metab. Eng. 2020;62:198-206. doi:10.1016/j.ymben.2020.09.007
- Zahoor A, Messerschmidt K, Helber SB, Gotsmy M, Peng B, Bleisch R, Oreb M, Toya Y, Jansen K, Boles E, Sauer M, Zanghellini J, Nidetzky B. ATPase-based implementation of enforced ATP wasting in Saccharomyces cerevisiae for improved ethanol production. Biotechnol. Biofuels 2020;13:185. doi:10.1186/s13068-020-01822-9
- Ye Z, Li S, Hennigan JN, Lebeau J, Bhatt S, Bhatt AP, Lynch JP, Bhatt S. Two-stage dynamic deregulation of metabolism improves process robustness and scalability in engineered E. coli. Metab. Eng. 2021;68:106-118. doi:10.1016/j.ymben.2021.09.009