Figure 5.

Example of discrete BN structure learned from 1600 variables. Dense variable cluster is centered around “identity” variable. This variable with 59 possible states was responsible for class prediction. All other nodes represent m/z values of chromatographic peaks with edges representing conditional dependency between variables. Blue nodes highlight structure motif identical for each network learned in CV. This network was also saved in.sif format (can be found in our repository by the link given below).