The VirHostScape user guide¶
I you are using VirHostNet 2.0 in your publication, please cite:
VirHostNet: a knowledge base for the management and the analysis of proteome-wide virus-host interaction networks.
Navratil et al. Nucleic Acids Research. Database issue 2009. [pubmed]
- The VirHostScape user guide
- 1. Short Presentation.
- 2. The VirHostScape "home page".
- 3. The VirHostScape "Visualisation" page.
- 4. Case study.
1. Short Presentation.¶
The VirHostScape web interface was designed to rapidly reconstruct (in few seconds depending on the complexity of the request) a fully customised protein interaction network (pin) representation around a protein (or a set of proteins), a domain, a pathway, a species or a publication of interest.
The VirHostScape web interface is organised in two pages:
- the "home page" provides basic tools ("Search" Panel, "Blast" panel, "Browse" panel) to interogate the VirHostNet 2.0 virus/host protein-protein interaction resource.
- the "visualisation page" is based on [cytoscapeweb.js] http://cytoscapeweb.cytoscape.org/ library (Version 1.0.4 14 June 2013) and provides tools to visualise and navigate throughout the virus/host protein interaction network.
2. The VirHostScape "home page".¶
The users can access the VirHostNet resource by using: 1) simple or multiple quick-search query, 2) a BLAST search and 3) by browsing the taxonomic summary.
2.1. The "Search" Panel.¶
The users can either make simple or multiple search according to the type of request (multiple search are only available for protein and publication queries):
Table: Summary of the different types of quick-search that could be performed over the VirHostNet 2.0 knowledgebase.
|Protein (Simple)||+||-||Retrieve all interacting partners of a selected protein.|
|Protein (Multiple)||-||+||Retrieve all interactions between a set of selected proteins.|
|Domain (Simple)||-||+||Retrieve all interactions between a set of proteins annotated with the domain.|
|Taxonomy (Simple)||+||-||Retrieve all interactions between a set of proteins annotated for the given species.|
|Publication (Simple)||-||+||Retrieve all interactions published in the selected article.|
|Publication (Multiple)||-||+||Retrieve all interactions published in the set of selected articles.|
- Search by protein accession number or name (UniProt and NCBI/REFSEQ primary accession number and name).
"I want to visualise all the protein partners of the STAT1_HUMAN protein."
- Search by Domain accession number (interpro - http://www.ebi.ac.uk/interpro/, supfam - supfam.org, pfam http://pfam.xfam.org, smart http://smart.embl.de/).
"I want to visualise all the interactions between the proteins annotated with the IPR001723 domain"
- Search by Pathway accession number (kegg pathway)
"I want to visualise interactions between proteins annotated as participating to the Jak-STAT signaling pathway"
- Search by Publication id (pubmed PMID)
"I want to visualise the interactions published with the PMID 18985028"
2.2. The "BLAST" Panel.¶
The users can search our database of interacting protein sequences (DIPS) for homologous protein sequences using a set of query protein sequences of interest. For more information about BLAST please visit their home page (http://blast.ncbi.nlm.nih.go).
2.3. The "Browse" Panel.¶Here, the users can explore and visualise the virus/host and virus/virus ppi content of the VirHostNet release 2.0 summarised at different taxonomic level:
- Baltimore group, (for more information about Baltimore classification)
- Family rank and
- Species rank.
3. The VirHostScape "Visualisation" page.¶
Prerequisite: The visualisation page relies on the Flash Player. If not available on your web browser, you can download and install it manually from the Adode website.
The graph representation (content, selection, colour) of the protein-protein interactions (left-panel) is synchronised with the tabular view (right panel) listing all proteins and protein-protein interactions (non-redundant and redundant).
The "button panel" at the top-left allows the users to select nodes of interest (all nodes or the neighborhood of a node) and to explore the direct partners of the selection (viral and/or host) or their connections (i.e. interactions between the selected protein).
The quick-search area at the top-right allows to make new search (see section 2.1) or to add results of new search to the current view using different entry points (protein accession number, domain, pathway, taxonomy, publciations).
3.1. Nodes (i.e. proteins)¶
3.1.1. The colour of nodes.¶
Nodes (and their borders) of the graph (i.e. the proteins) are coloured by default according to the UniProtkb and the RefSeq division (human division=blue; viral division=red; bacteria division=green; etc.).
The legend of colours appears in the "Otions>Colorize>Node" section and can be personalised. Here nodes (and their borders) can be colourized accrording to taxonomic level provided by the NCBI (Species rank, Genus rank, Familiy rank, Third level of taxonomic rank or Baltimore group for viruses).
3.1.2. The size of nodes.¶
The nodes are resized by default according to the node degree (ktot) (i.e. the number of partners) computed in the whole network model (built with VH, VV, HH protein-protein interactions). The size of the nodes is proportional to the degree value (larger nodes having more partner than smaller one).
In "Options>Resize", the size can be modified according to different properties (degree, betweenness, transitivity, closeness) computed in three types of networks : 1) the whole network model, 2) the host network model (built with HH protein-protein interactions) 3) the virus/host network model (built with VH protein-protein interactions).
3.1.3. Filtering nodes.¶
In "Options>Filters>Nodes", the nodes can be filtered according to the network properties using a dedicated slider. Hence the users can remove nodes above and/or below threshold values.
3.2. Edges (i.e. protein-protein interactions)¶
3.2.1. The colour of edges.¶
Edges of the graph (i.e. interactions) are coloured by default according to the UniProtkb and the RefSeq division of the interacting partners.
Intraspecies interactions are coloured according to the division of the sepcies (human-human=blue; viral-viral=red; bacteria-bacteria=green; etc.).
Interspecies interaction annotated as Virus/Host species relationships in UniProt and Viralzone are coloured in red and those not annotated are coloured in grey. By this way users can easily differentiate ppi supporting a true virus/host relationship from more artificial ones, related to more technical considerations.
The legend of colours appears in the "Otions>Colorize>Edges" section and can be personalised.
3.2.2. The width of edges.¶
The edges are resized by default according to the number of indepedant experiments supporting the non-redundant protein-protein interactions. The width of the edges is proportional to the number of indepedant experiments.
In "Options>Resize", the width of edges can be modified according to different properties (number of independant experiments, number of different methods, number of pmid, number of database).
3.2.3. Filtering edges.¶
In "Options>Filters>Edges", the edges can be filtered according to the level of support of the interaction (number of independant experiments, number of different methods, number of pmid, number of database) using a dedicated slider. Hence the users can remove edges above and/or below threshold values. Edges can also be filtered according to the database referencing the interactions in "Options>Filters>Database".
4. Case study.¶
Case study 1. Infer the EBOV (Zaire ebolavirus isolate Ebola virus H.sapiens-wt/SLE/2014/ManoRiver-G38160)protein interaction network¶
The goal of this case study is to illustrate how the VirHostNet 2.0 knowledgebase can be used to infer a putative virus/human protein interactions from an annotated genomic sequence.
We focus this case study on the genome of the ebola virus (EBOV) isolated from patients during the recent outbreak in Sierra Leone (West Africa, May 2014). See here the alert raised by the World Health Organization on May 25, 2014.%
Step 1: Paste the set of annotated protein sequences (download) of the KM233090.1 genome in the text-are or load the sequences from a file. Then run BLASTP with default paremeter values.
Step 2: Select the BLASTP hits from the table of result. Then search for host and/or viral partners of the selected hits, to infer new protein-protein interactions based on interology assumption. BLASTP hits are represented by diamond shape in the graph and flagged with yellow stars in the table. Mouse-over the yellow stars shows the name of the original query sequence.
Note that the best homologous candidates are automatically selected based on the following criteria 1) the best e-value, 2) the best query and subject alignment coverage (here 100%).
Step 3: Search for interactions between the set of candidate human proteins predicted to interact with the EBOV proteins.