Skip to main content
. 2020 Jun 11;11:1168. doi: 10.3389/fimmu.2020.01168

Figure 3.

Figure 3

Identifying sources of error. Deviation from observed behavior for (A) the 82 candidate models with <5% overall error was decomposed into deviation across markers at each of the 10 experimental conditions used to constrain model identification (left panel) and into deviation across all conditions for constrained markers during acute AOM (right panel). A similar decomposition of error was presented for (B) the 44 candidate models remaining after validation against new experimental measurements. Departure indicates the magnitude and direction (positive or negative) of the error contributed by each network entity. Error is computed as the ratio of discrete deviations across all markers to the overall maximum possible deviation.