Fig. 10

Creating a Bayesian network from an annotated map of metabolism. Considering a An annotated map of metabolism with enzymes named E1, E2, E3 and E4 and a SOM score associated to each enzyme for each reactions. b Illustrate the application of the rule: “only one enzyme can catalyze a reaction”. There is a reaction from A to C, divided in three reactions one for each enzyme annotating the reaction. Those reactions need to be reduced in one with a SOM score used as probability for the Bayesian Network. SOM score annotating reactions are indicators of enzymatic affinity so the reaction with maximal SOM score is selected. Here is the reaction annotated by E3. c Illustrate a different case where there are reactions from D and from C leading to production of F. This is the application of “if an enzyme is recruited by a reaction, another enzyme has to be recruited for a second reaction with the same product”. Here the maximal SOM score annotating reactions is 0.77 for the reaction from D to F annotated by E1. Because this reaction is the retain reaction from D to F, E1 is no more allowed to annotate the reduced reaction from C to F. The reaction retained from C to F, is the second reaction with the maximal SOM score which is the reaction annotated by E2. d Illustrate the final reduced graph used as Bayesian Network to calculate production probability score of metabolites