Fig. 2: Output of AutoCD for the two-strain stable steady state objective.

a Dendrogram generated by hierarchical clustering of the adjacency matrices for each model in the two-strain model space. All possible interactions are shown in the network illustration. Each column of the heatmap represents a possible interaction between state variables, where green indicates the interaction exists for the model and black indicates absence of the interaction. The bar chart shows the mean posterior probability of each model. The scatter points indicate the posterior probability of each replicate and the error bars indicate the standard deviation. n = 3, where each replicate consists of 180,000 accepted particles. b Shows the mean posterior probability of three models with the highest posterior probability when subsetted for number of parts expressed in order of increasing complexity (2, 3 and 4 parts). The scatter points indicate the posterior probability of each replicate and the error bars indicate the standard deviation. n = 3, where each replicate consists of 180,000 accepted particles. Bayes factors (BF) are shown for pairwise comparison of the three models. Model schematics show the interactions between strains (blue and green), bacteriocin (red) and QS molecules (purple). c Posterior parameter distributions of several tunable parameters in m₆₆ and m₆₂. The top and left plots show 1D density distributions of each parameter, central distributions are 2D density distributions for each pair of parameters. Pearson correlation coefficients are shown on the top side of the diagonal for each parameter pair.