Modeling SCOBY Biofilm Growth in Kombucha

ISEF Category: Computational Biology and Bioinformatics

Subcategory: Computational Biomodeling  ·  Difficulty: Intermediate  ·  Setup: Home Setup  ·  Time: 1 to 2 Months

The Hook

A SCOBY can grow in layered patterns that look random, but they often follow simple local rules. That makes kombucha biofilm a good fit for simulation. You can test whether sugar and oxygen gradients explain what your camera sees. This project connects coding, image analysis, and real biology.

What Is It?

A SCOBY is the cellulose-rich biofilm that forms during kombucha fermentation. A biofilm is a community of microbes stuck together in a slimy matrix. Think of it like a city built by tiny workers, where each cell responds to nearby conditions instead of a master plan.

An agent-based model lets you simulate those tiny decisions. Each agent can stand in for a microbe or a patch of biofilm. You set simple rules, like how cells grow, spread, or cluster when sugar or oxygen changes. Then you compare the simulation to time-lapse photos of real SCOBY growth and ask whether the same rules can explain both.

Why This Is a Good Topic

This is a strong science fair topic because you can test a real biological pattern with code, images, and clear variables. Sugar and oxygen gradients affect fermentation, so your model has a real physical basis. You can measure growth area, thickness proxies, edge shape, and pattern change from photos without needing a wet lab full of advanced gear. You also get to learn modeling, image analysis, and experimental design in one project.

Research Questions

• How does sugar concentration affect the growth rate and surface coverage of a SCOBY biofilm?
• What is the effect of oxygen exposure on the edge shape and layering pattern of a growing SCOBY?
• Does a simple agent-based model reproduce the observed radial growth pattern of home-grown SCOBYs?
• To what extent do gradient-based rules improve model fit compared with a uniform-growth model?
• Which parameter settings best match the time-lapse image sequence of a real SCOBY?
• How does the initial inoculum size affect early clustering and final biofilm area?

Basic Materials

• Wide-mouth glass jar or clear container.
• Black-and-white background sheet for photos.
• Smartphone with a consistent camera mode.
• Tripod or phone stand.
• Ruler or printed scale marker for images.
• Household sugar.
• Tea for kombucha starter medium.
• Kombucha starter liquid with a SCOBY.
• Gloves and basic food-safe handling supplies.
• Notebook or spreadsheet for observations.

Advanced Materials

• Computer with Python or Mesa installed.
• NetLogo software.
• ImageJ for image measurements.
• Digital scale with 0.1 g accuracy.
• Light meter or phone lux app for monitoring photo conditions.
• Clear acrylic or glass test containers for repeatable geometry.
• Access to a temperature-controlled space.
• R statistical software for model comparison and plotting.
• Optional USB microscope or macro lens for edge detail.

Software & Tools

• Mesa: Runs Python agent-based models for local growth rules and gradient effects.
• NetLogo: Lets you prototype agent-based biofilm behavior with visual sliders and patches.
• ImageJ: Measures SCOBY area, edge roughness, and frame-to-frame change from photos.
• Python: Handles data cleaning, simulation, and graphing for model calibration.
• R: Compares model fits and builds statistical plots for growth trends.

Experiment Steps

1. Define the biological pattern you want to explain, such as radial spread, edge roughness, or layering density.
2. Choose one environmental variable to change first, like sugar level or oxygen access.
3. Design a photo plan that keeps framing, lighting, and scale consistent across the whole growth period.
4. Build a simple agent-based rule set that turns local growth decisions into a visible biofilm pattern.
5. Set up a way to compare model output with real SCOBY images, using the same measurement method for both.
6. Plan a test of model fit, then decide which parameter changes count as a better explanation.

Common Pitfalls

• Changing the camera angle between photos, which breaks your ability to measure growth over time.
• Using uneven lighting, which creates fake texture changes in the biofilm images.
• Comparing model output to raw photos without scaling, which makes area and shape measurements inconsistent.
• Tracking too many biological rules at once, which makes it hard to know which parameter caused the pattern.
• Ignoring batch-to-batch variation in SCOBY starting material, which can hide the effect of sugar or oxygen.

What Makes This Competitive

A stronger project does more than draw a pretty simulation. It tests whether your model predicts real image data better than a simple baseline, and it explains why. You can raise the level by comparing multiple gradient rules, running repeated trials, and using a clear error metric. If your analysis connects pattern formation to a biologically plausible mechanism, the project gets much stronger.

Project Variations

• Model how oxygen exposure at the jar surface changes SCOBY edge growth and thickness over time.
• Compare a uniform-growth agent model with a gradient-driven model using the same time-lapse photo set.
• Test whether different starter sources produce distinct biofilm shapes, then see which model parameters best match each source.

Learn More

• NetLogo Models Library: Search the official NetLogo site for sample agent-based models and documentation on building patch and agent rules.
• Mesa Documentation: Find the Python Mesa project docs for tutorials on agent-based modeling and model structure.
• ImageJ User Guide: Use the official ImageJ documentation to learn how to measure area, edges, and grayscale features in photos.
• NCBI Bookshelf: Search for free chapters on biofilms and microbial communities in publicly available textbooks.
• PubMed: Search review articles on kombucha microbiology, cellulose biofilms, and gradient-driven biofilm growth.
• MIT OpenCourseWare: Search for free course notes on modeling, simulation, and biological systems dynamics.