Virus–virus and virus–host protein–protein interaction methods summary. (A) Distribution of protein–protein interactions according to experimental procedures (yeast-two hybrid, co-immunoprecipitation, pull-down and others). (B) Distribution of protein–protein interactions identified by one, two or more distinct methods.

The VirHostNet Query interface. (A) Menu of the VirHostNet interface. Users are allowed to search for protein–protein interactions according to gene, protein, biological function (GO), pathway (KEGG), protein domain (INTERPRO) or PMID information. For example, in the ‘Pathway’ query interface of the ‘Query’ menu, users can enter ‘mTOR’, then select the right pathway defined in KEGG database and ask for proteins annotated in this pathway (see result in Figure 4B). (B) List of the 49 proteins belonging to the mTOR pathway. The ‘?’ link at the end of each protein line allows users to access functional annotations of the protein in external databases. (C) Protein–protein interactions research panel. Users can select one or more proteins from the list of proteins (by default, all proteins are selected). Then, from the preferences panel, they have to choose the kind of interaction to retrieve (virus–virus, virus–host or host–host) or the database origin. Finally, they have to choose an operation (‘neighbours’ or ‘sub-graph’). The ‘neighbours’ button allows users to retrieve all direct protein–protein interactions from a single protein or a list of selected proteins. The ‘sub-graph’ button allows users to retrieve only protein–protein interactions occurring between selected proteins (see result in Figure 4D). (D) List of the 89 protein–protein interactions occurring between proteins of the mTOR pathway. For the interacting proteins, NCBI taxonomy name, ENSEMBL gene acc (host)—NCBI protein acc (viruses) and official gene name (host)—product name (viruses) are given. The ‘?’ link at the end of each interaction line allows users to access interaction annotations (Figure 4E). In the protein–protein interactions research panel, in addition to ‘neighbours’ and ‘sub-graph’ buttons, the ‘visualize’ button allows from a list of interactions the graphical visualization of the resulting network (see result in Figure 5). (E) More detailed information concerning protein–protein interactions are available. PMID, PSI-MI method accession number and database origin of the selected interaction are given.

Flow chart of the VirHostNet knowledge-based system integration process. Virus–virus, virus–host and host–host protein–protein interactions were integrated from 10 external public databases. BLAST and IPI cross-reference databases were used to link protein interactor identification numbers to NCBI (viruses) and ENSEMBL protein accession numbers (e!) (host) databases respectively. Virus–Virus and Virus–Host protein–protein interactions were also extracted from literature (Literature Curated Interactions) and curated from databases (Database Curated Interactions). Functional annotations of cellular proteins were extracted from Gene Ontology, KEGG and INTERPRO databases.

VirHostNet protein–protein interaction content. (A) In each Baltimore group (dsDNA, ssDNA, dsRNA, ssRNA−, ssRNA+, retro transcribed), for each viral family, the number of virus–host protein–protein interactions (ppi) and the associated number of viral (v) and host cellular protein (h) interactors, the number of virus–virus protein–protein interactions (ppi) and the associated number of viral (v) interactors are given. (B) Distribution of virus–virus interactions according to viral Baltimore groups. (C) Distribution of virus–host interactions according to viral Baltimore groups.

VirHostNet Visualisator. Graph representation of the mTOR-infection network is drawn (right of the applet). Only interacting proteins of the mTOR pathway are represented (cellular proteins = blue nodes; host–host protein–protein interactions = blue edges). Viral interacting proteins are also added to define the infection network (viral proteins = coloured nodes according to viral taxonomy; virus–host protein–protein interactions = red edges). Cellular proteins of the network targeted by at least one viral protein (red stroke) are highlighted. The width of the edges is roughly proportional to the number of PMIDs used to describe the interaction. The width of the nodes is roughly proportional to the cellular degree, i.e. the number of cellular partners in the whole network. The taxonomical classification of viral proteins interacting with the mTOR network is given (left panel of the applet). The menu of the applet (top of the applet) allows users to define nodes and edges preferences (width, colour, labelling), interaction filtering (according to the number of methods, number of PMIDs or both), network navigation (expand viral or host neighbours) and graph layout. To obtain this figure, from the list of the 89 selected protein–protein interactions obtained in Figure 4E, users have to click on ‘visualize’ button. Then from the applet, in picking mode, they have to select all proteins and ask for viral proteins neighbours (in expand menu).