First, it has been studied and validated in great detail [26—28] , such that the available signaling network from the KEGG database [22] can be used as a reliable source to compare to.

Nodes of KEGG pathways can contain several genes.

We selected genes such that they are all assigned to different KEGG nodes using a weighted maximum bipartite matching of low off-target siRNAs and unique KEGG nodes.

Comparing the learned network to the known KEGG pathway of the genes shows that our method recovers networks significantly better than classical nested effects models without capturing of hidden signaling.

Before comparing a network to the corresponding KEGG pathway, we also built its transitive closure.

We generated 30 network structures with 5 signaling genes, randomly sampled from KEGG pathway maps [22] as previously described in [6].

We performed GO and KEGG enrichment analyses in R using the Bioconductor packages Bio-Mart and KEGGgraph, respectively [43, 44].

For the KEGG enrichment analysis, as the universe of human proteins, we used the human proteins available in KEGGgraph that participated in at least one KEGG pathway [44].

1 to identify biological pathways targeted by all tested B. mallei proteins using the Kyoto Encyclopedia of Genes and Genomes ( KEGG ) annotation database [32].

Gene Ontology (GO) terms and Kyoto Encyclopedia of Genes and Genomes ( KEGG ) pathways statistically significantly enriched in human proteins interacting with known and/ or putative B. mallei Virulence factors.

Functional analysis of the respective genetic signatures implicates a wide spectrum of Gene Ontology terms and KEGG pathways with condition-specific information content, including iron transport, transferases, and enterobactin synthesis.

Various annotations including Gene Ontology terms, KEGG pathways, and InterPro protein domains are investigated.

Pathway diagrams were re-plotted from the KEGG database [74].