Find the Optimal Medium with Constellab Bioprocess | Use case

Optimizing a culture medium is one of the most time-consuming challenges in bioprocess development. You run dozens of fermentations, collect a lot of time-series data, and then spend weeks trying to figure out which ingredient actually made the difference.


We built the Constellab Bioprocess Dashboard to change that.


The Use Case


Inspired by published research on Lactobacillus delbrueckii subsp. bulgaricus we simulated a Design of Experiments with 6 culture media and 120 fermentation runs. The media differed in four mineral components: dipotassium phosphate (K₂HPO₄), magnesium sulfate (MgSO₄), manganese sulfate (MnSO₄), and ammonium citrate.


The question: which ingredients drive growth — and which ones hold it back?



YouTube video

What Happened


We uploaded four files — a sample sheet, time-series measurements, a medium composition table, and online OD monitoring data — and let the dashboard do the work.


In a few clicks, the dashboard:


  • Visualized all 120 growth curves side by side, colored by medium
    • Extracted kinetic parameters from each curve using a logistic growth model — maximum biomass (A), growth rate (μ), and lag time
      • Ran a PCA on medium compositions alone
        • Identified the key drivers using PLS Regression and Random Forest: K₂HPO₄ and MgSO₄ have a strong positive effect on growth; ammonium citrate is an inhibitor
          • Revealed a hidden confound: Medium M4 has the same high K₂HPO₄ as the best-performing medium M3 — but its growth is poor because high citrate cancels the benefit. Without multivariate analysis, this would have been easy to miss.
            • Suggested the optimal medium through evolutionary optimization: maximize K₂HPO₄ and MgSO₄, keep citrate low — exactly matching the best medium in the dataset.

              Try It Yourself


              We've made the input files available so you can reproduce this analysis on your own Constellab instance. Follow along with the video walkthrough, or use it as a template for your own fermentation data.


              • Download the input files









                Dataset inspired by: Tovilla-Coutino et al., 2022 — "Effect of fermentation pH, temperature, and harvest on cell growth and functional properties of frozen and freeze-dried Lactobacillus delbrueckii subsp. bulgaricus CFL1", Recherche Data Gouv. https://doi.org/10.15454/FZHIE0


                Comments (0)

                Write a comment
                Shine Logo
                Stories & feedback

                Do you have a resource to share?

                Share it to accelerate projects for the entire community.