What is DifferentialNet?
Why use DifferentialNet?
How do you know if a protein-protein interaction is differential in a tissue?
Where do you get the protein-protein interactions data from?
Where do you get the RNA-Sequencing data from?
Which proteins are highlighted and why?
Which interactions are colored and why?
Sometime you say gene and sometimes a protein, do you mean a gene or a protein? Do you support alternative spliced proteins?
Are interactions in the lower percentiles expressed in the selected tissue?
How do you know that the differential score computation method captures tissue aspect better then static interactions?
How can I select proteins and activate the mouse right-click menu?
How can I access your database programmatically?
How can I contact you?
For additional information:
To go back to the DifferentialNet home page click here.
Contact us at: Esti Yeger-Lotem at estiyl@bgu.ac.il,Omer Basha atnetbio@omerbasha.com
DifferentialNet is a novel database that provides users with differential interactomesof human tissues. Users query DifferentialNet by protein, and retrieve its differential protein-protein interactions (PPIs) per tissue. By this, users can filter out PPIs that are relatively stable across tissues, and highlight PPIs that show relative changes across tissues. To compute differential PPIs, we integrated available data of experimentally-detected PPIs with RNA-sequencing profiles of tens of human tissues gathered by the Genotype-Tissue Expression consortium (GTEx) and by the Human Protein Atlas (HPA). We associated each PPI with a score that reflects whether its corresponding genes were expressed similarly across tissues, or were up- or down-regulated in the selected tissue relative to other tissues. The differential PPIs can be used to identify tissue-specific processes and to decipher tissue-specific phenotypes. Moreover, they unravel processes that are tissue-wide yet tailored to the specific demands of each tissue.
Large-scale protein-protein interaction (PPI) detection screens tend to provide limited physiological context. DifferentialNet enables users to evaluate the PPIs of query proteins across tissues, such that PPIs that are relatively stable across tissues are filtered out while PPIs that show relative changes across tissues are highlighted. Users can find, for example, the possible interactions of a query protein in heart compared to all other tissues, find interaction partners that are common across all tissues (not differential), or find the tissues where a specific interaction is up- or down-regulated. The differential PPIs can be used to identify tissue-specific protein functions, and to unravel processes that are tissue-wide yet tailored to the specific demands of each tissue. DifferentialNet supports 42 different human tissues from the GTEx project, and 27 human tissues from the HPA project using RNA-sequencing data.
How do you know if a protein-protein interaction is differential in a tissue?
DifferentialNet associate a PPI with a score that represents whether its corresponding genes were expressed similarly across tissues, or were up- or down-regulated in the selected tissue. The association does not guarantee that the interaction occurs, instead it implies that the interaction is more likely or less likely in a specific tissue relative to all other tissues.
Where do you get the protein-protein interaction data?
Experimentally detected protein-protein interactions were obtained from four major databases: BioGRID , DIP, IntAct, and MINT. The extraction was done using the MyProteinNet web-tool [3], which extracts only interactions that were detected by established methods for identifying physical interactions between proteins (e.g., co-localization was excluded). Details regarding a PPI can be viewed in the Properties tab upon selecting the PPI in the network.
Where do you get the expression data?
We extracted RNA-sequencing profiles from two leading sources: the Genome-Tissue Expression (GTEx) consortium [1]and the Human Protein Atlas (HPA) [2]. GTEx data included 421 RNA-sequencing samples from 42 tissues, and HPA data included 192 RNA-sequencing samples from 27 tissues.
Which proteins are highlighted and why?
Query proteins are diamond-shaped. Proteins with an octagon shape are related to human diseases according to OMIM, and were downloaded from the Ensembl BioMart web-service.
Which interactions are colored and why?
DifferentialNet colors interactions by their score percentiles. If the interaction has a positive score, it means the interaction is over expressed in the query tissue relative to all other tissues and will appear in a shade of red by its score percentile. If the interaction has a negative score, it means the interaction is under expressed in the query tissue relative to all other tissues and will appear in a shade of blue by its score percentile.
Sometime you say gene and sometimes a protein, do you mean a gene or a protein? Do you support alternative spliced proteins?
Most of the available protein-protein interactions data are oblivious to alternatively-spliced isoforms. Therefore, DifferentialNet associates each protein-coding gene with a single protein product. For simplicity, we refer to the gene and its protein product interchangeably.
Are interactions in the lower percentiles expressed in the selected tissue?
Interactions in the lower percentiles can be simply down-regulated in the presented tissue relative to other tissues, or might be missing from that tissue in case one of the proteins is not expressed. To find out if the interacting proteins are expressed in a specific tissue we recommend using our TissueNet v.2database.
How do you know that the differential score computation method captures tissue aspects better then static interactions?
We tested extensively whether the differential interaction scoring scheme captures tissue biology. In the test described below, we used GO enrichments. For this, we combined text mining and manual curation to associate 5,318 GO process terms with 26 main tissues (for example, we associated aortic valve morphogenesis with heart). Then, we created a ranked gene list per tissue, where genes were ranked by the score of their most up-regulated interaction in that tissue. We then carried a GO term enrichment test for this ranked list (via the GOrilla web-server). Next, we tested whether the GO terms enriched in this ranked list (adjusted p-value < 0.001) overlapped significantly with GO terms that we associated with the studied tissue, by using Fisher exact test (Figure 1A). In 35/42 tissues and sub-tissues that we tested, the overlap was significant. To further test the tissue-specificity of the differential networks, we tested whether the GO terms enriched in one tissue also overlapped significantly with GO terms that we associated with another tissue (Figure 1B). We found that enrichment was typically most significant for the GO terms associated with the studied tissue. In case of closely-related tissues, such as brain and nerve, or muscle and heart, we also observed enriched for the other tissues’ GO terms. Thus, differential interactions appear to capture biological differences between tissues.
How can I select proteins and activate the mouse right-click menu?
To select proteins and interactions you can left-click over a protein or interaction in the network, hold the shift key to select multiple items, or hold the shift key and mark an area by left-clicking and dragging the mouse. Once you selected them you can (i) view their properties in the Properties tab, (ii) view their Gene Ontology [4]annotations, (iii) view differential scores in all tissues for selected interactions in the Tissues tab or (iv) right-click the mouse to obtain the mouse right-click menu.
How can I access DifferentialNet database programmatically?
We offer a web-service method to query the database using the same parameters as in the web-UI (dataset, tissue, threshold, gene). The method is in a REST format and can be accesses from all programing languages by using an http request, a wget command on linux and windows-based computers, or from any web-browser. The method returns a JSON object that can be parsed by any programming language, to obtain a network object of the query output. In addition, we offer a download option that provides the complete database in CSV format.
For more information please see our web-service documentation page for specifications of this method (http://netbio.bgu.ac.il/diffnet-api).
Please email Esti Yeger-Lotem at estiyl@bgu.ac.ilor Omer Basha atnetbio@omerbasha.com
We would love to hear from you!
References
- GTEx Consortium, The Genotype-Tissue Expression (GTEx) pilot analysis: multitissue gene regulation in humans.Science, 2015. 348(6235): p. 648-60.
- Uhlen, M., et al., Proteomics. Tissue-based map of the human proteome.Science, 2015. 347(6220): p. 1260419.
- Basha, O., et al., MyProteinNet: build up-to-date protein interaction networks for organisms, tissues and user-defined contexts.Nucleic Acids Res, 2015. 43(W1): p. W258-63.
- Ashburner, M., et al., Gene ontology: tool for the unification of biology. The Gene Ontology Consortium.Nat Genet, 2000. 25(1): p. 25-9.