Hybrid cellular automaton-based model for quorum sensing-controlled biofilm evolution

Computational modeling and in silico studies are critical for understanding how the spatial organization of biofilms contributes to antimicrobial tolerance and persistence. The paper presents a novel hybrid computational framework for the discrete-in-space dynamical modeling of bacterial biofilms. The approach combines a cellular automaton, which generates naturalistic biofilm morphology on a hexagonal lattice, with discrete analogues of reaction-diffusion equations governing the distribution of nutrients and signaling molecules. This design incorporates a quorum sensing feedback mechanism that links local signaling molecule concentration to biofilm spreading. The simulation system was developed in C# on the Unity platform, and the source code together with a Windows executable release was deposited on Zenodo. The biological plausibility of the simulated AHL and population dynamics was assessed through a semi-qualitative comparison with published experimental observations. The results reproduce distinct growth regimes ranging from sparse, branched colonies to compact biofilms with continuous fronts. A two-parameter analysis reveals a curved transition boundary in the nutrient-threshold plane, demonstrating that the effective quorum sensing activation threshold depends on nutrient availability. The qualitative comparison shows that the model can generate a biologically plausible transient AHL profile, including signal accumulation, formation of a maximum, and subsequent decrease, together with saturating population dynamics. This comparison is not intended as quantitative validation of absolute timing, concentration, or the detailed biochemical mechanism of AHL removal. These results support the proposed approach as a mechanistic tool for studying how quorum sensing and nutrient limitation jointly shape biofilm morphology. By providing an interpretable mechanistic simulation framework with explicit state variables, transition operators, and experimentally comparable outputs, the model establishes a basis for future AI-assisted workflows for diffusion-solver acceleration and automated parameter calibration.