Fig. 4.

Metabolic network inhibition of hydrazine learned by abductive SLPs from probabilistic examples (SLP_P). Each observed metabolite is associated with its concentration and the estimated empirical probability. The learned posterior probabilities for each inhibition (in two directions) are shown in the associated ellipse. For example, the left corner ellipse specifies a learned inhibition in the form of two SLP clauses: ‘0.000988:inhibited(2.6.1.39,2-og,l-2-aa,t).’ and ‘0.0592:inhibited(2.6.1.39,l-2-aa,2-og,t).’, which mean there is a significant inhibition from metabolite l-2-aa to metabolite 2-og with a probability 0.0592.