Bacterial Growth and Antibiotic MIC Lab (3D)

A Blender bench with a broth flask and an agar plate: Monod growth with lag, death phase, antibiotic kill curves around the MIC and a disc-diffusion inhibition zone.

Switch to 2D

Presets

Playback

Culture

Antibiotic and agar disc

View

Camera

Layers

Quality

Model check

The science

What you see. A batch culture of a generic, harmless laboratory rod grows in a flask of broth. The curves, the cloudiness of the flask, the cells under the microscope and the lawn on the agar plate all come from one set of equations. Nothing here uses a dangerous organism, and the model gives no guidance for handling pathogens.

Growth with a lag (Baranyi-Roberts form). The population N grows at g(t) = muT S/(Ks + S) alpha(t), where S/(Ks + S) is the Monod saturation in the nutrient S and alpha = q/(1 + q) is the fraction of cells that are "ready" to divide. The readiness variable obeys dq/dt = muT q, and q0 = 1/(exp(muT lambda) - 1) gives a chosen lag time lambda. After the lag the rate approaches the Monod value: exponential growth, then deceleration as S falls.

Temperature. muT follows the Ratkowsky square-root law, sqrt(muT) = b (T - Tmin)(1 - exp(c (T - Tmax))), with illustrative constants (Tmin = 3 C, Tmax = 48 C, c = 0.25 /C, b = 0.0433) that give a doubling time of about 22 minutes at 37 C. Real strains have their own constants.

Nutrient and death. dS/dt = - g Ngrowing / Y', with Y' = 109 cells/mL per g/L (about 1012 cells per gram, a typical order of magnitude). When the nutrient is gone, cells die at kd = kd0 Ks/(Ks + S). Dead cells stay in the flask, so turbidity (OD600, about 8 x 108 cells/mL per unit) stays high while live counts (CFU) fall. The phase label (lag, exponential, stationary, death, antibiotic kill) is read from the live net rate d ln N/dt, the lag state alpha and the nutrient history.

Antibiotic and the MIC. The kill rate follows the pharmacodynamic form of Regoes et al. (2004): K(C) = (mu + psi)(C/MIC)kappa / ((C/MIC)kappa + psi/mu). In exponential growth the net rate mu - K(C) is exactly zero at C = MIC, which is how the MIC is defined here; psi is the largest kill rate beyond stasis and kappa the steepness. Killing is scaled by 0.25 + 0.75 g/mu, so slow-growing cells are killed more slowly (as for many growth-dependent drugs). A tolerant fraction fp of the inoculum does not grow and is not killed (non-growing cells die only slowly by starvation), which produces the textbook two-phase kill curve with a plateau. This is a tolerance illustration, not a model of genetic resistance.

Agar disc diffusion. The inhibition zone radius uses the Cooper-Woodman critical-time picture: r2 = 4 D tc ln(C0/Cc), with D = 5 x 10-6 cm2/s, tc = 4 h and the disc loading expressed as a multiple of the MIC. The zone grows during the first hours and then freezes while the lawn thickens outside it. Real zone sizes depend on agar depth, inoculum and drug chemistry; the formula is only a scaling.

Checks (run in the page). (1) With ample nutrient and no lag the fitted growth rate equals the Ratkowsky value (doubling-time check). (2) Mass balance: cells made equal yield times nutrient used. (3) At C = MIC the simulated net rate is zero, and at 8 x MIC the simulated rate equals the closed-form dose-response value.

Simplifications. Single species, well-mixed, no pH or oxygen limits, no maintenance energy, no cell-size dynamics (the rods only change length for display), no mutation or resistance evolution, drug concentration constant after it is added, and the constants are illustrative rather than fitted to one strain.

What to try

  • Run "Lab strain, 37 C" and read off lag, exponential, stationary and death phases; compare live CFU with turbidity in the first graph.
  • Lower the temperature to 10 C and compare the doubling time with the Ratkowsky curve.
  • Add the drug at the MIC: growth stalls. Add it at 4 x MIC and at 8 x MIC with a tolerant fraction: the kill curve bends into a plateau.
  • Raise the disc loading and watch the inhibition zone widen, as the square root of ln(loading).
Loading model...
Loading Blender model...
Growth curve with lag, exponential, stationary and death phases
Monod curve: growth rate against nutrient
Antibiotic dose-response: net growth against C/MIC
Time-kill curves at 0.5 to 8 x MIC
Made with 🟠 Blender · Three.js