Figure 4

Mixture Identifiability Smaller mixtures (e.g. two mechanisms) may be identifiable, but de novo identification of mixture mechanisms across larger numbers of mechanisms (e.g. five) or when the number of mechanisms is unknown (far right column) is not possible with our approach. Mixture Similarity Different mixtures of mechanisms can produce functionally equivalent systems, here quantified by normalized Ascendency²⁰. Pies are example networks generated by mixtures of the five mechanisms, grouped by their shared Ascendency. Function Identifiability Leave-one-out cross validation of system development (normalized Ascendency). Each point is a single mixture network, with randomly chosen proportions of the five mechanisms and governing parameters within each mechanism. R² = 0.67.