You are here: Home / TECHNOLOGY - abSYNTH / Pepper - Cytoscape app

Pepper - Cytoscape app

PEPPER (Protein complex Expansion using Protein-Protein intERactions) is a plugin which finds meaningful pathways / complexes connecting a protein set members within a PPI-network using multi-objective optimization.

Installation

System requirements

PEPPER is platform independent and runs on Java virtual machine within the Cytoscape environment. It requires minimum 512MB RAM but 2GB RAM is recommended for work with larger networks (over 100 nodes).

To run it requires Java version 6 or higher (either OpenJDK or Oracle Java SE) and Cytoscape version 3.0 and higher.

The tutorial hereinafter developed is suited for PEPPER 1.4.+ versions.

How to enable PEPPER in Cytoscape?

Open Application Manager in Cytoscape (Apps -> App Manager) and select PEPPER from the list of plugins.

Cytoscape_App_Manager

To start PEPPER choose analysis from Apps -> PEPPER menu or directly click on the PEPPER tab in Cytoscape Control Panel (left-hand side panel).

Analysis

Input

  1. choose Pepper in the control panel
  2. select a chassis organism among the list of available species (default is H. sapiens)
  3. select the appropriate identifier format for your protein set (default is HGNC)
  4. select molecular interaction network (default is Hippie for H. sapiens)
  5. Input your bait protein name for AP/MS experiments (ex.: this baitOR uncheck the box to perform an analysis on another kind of data
  6. Input the rest of your protein set (one protein ID per line, ex.: these preys) OR Insert your proteins list from a text file (one protein ID per line) by clicking on "Choose..." OR Select nodes from an existing Cytoscape network view, then click on "Select"
  7. Optionally, adjust the genetic algorithm parameters
  8. Click on "Start" to start the analysis

PEPPER_input_panel

 

Sub-graph extraction results

The computation may take up to a few minutes. When results are ready, they will appear in Cytoscape Results Panel in the Main prediction results tab.

Sub_graph_extraction_results

Two main sub-graphs are generated alongside the Main prediction results panel:

  • Merge by modularity sub-graph
  • Union of all Pareto solutions sub-graph

Next to each graph miniature in the Results panel are generated their own properties in terms of modularity, coverage of the user input protein list, density and number of proteins added by the algorithm.

The All solutions button allows to display each of the Pareto solutions one by one to check the multi-objective algorithm output. Again, similar properties as the ones displayed for the Merge and Union sub-graphs are available.

Network view

Sub-graphs extracted by PEPPER are visualisable in Cytoscape Networks View Panel.

Merge_subgraph

Once results are computed a dialog window appears to explain network views legend.

Views_legend

Hexagonal red nodes correspond to proteins that were added by PEPPER algorithm, rounded green nodes are proteins in the complex that were also in the use input proteins list whereas rounded light green nodes are proteins that were removed by PEPPER algorithm. Edges between nodes representing protein interactions are coloured in respect to the previous node colours.

Users can customise network views by hiding or showing proteins that were added by PEPPER algorithm respectively with a click on the Hide expansions button which then becomes a Show expansions button in the results panel corresponding to a specific network. A similar dynamical display is available to show or hide proteins that were removed by the algorithm by means of the Hide/Show missed proteins.

Post-processing results

Results concerning the post-processing pipeline applied after PEPPER algorithmic procedure is achieved are visualisable in Cytoscape Results panel in the Post-processing tab. They display three types of analysis:

  • Topology properties of the extracted sub-graphs. Four distinct properties (namely Degree, Clustering coefficient, Modularity and Closeness centrality) are computed for each sub-graph. Merge and Union graph gains in each of those properties are visualisable as bar plots in the Topology tab.

Topology_subgraphs

Bar plots display differences in sub-graph connectivity features while considering a graph without or with PEPPER protein expansions (respectively Initial and Expansion labels). Bar plots grouping all solutions including Pareto solutions are able to be generated by clicking on a specific topological property in the Select... drop-down list. Row data is also visualisable in a dialog window by clicking on the Display row data button.

  • Known complexes comparison which displays reference protein complexes from golden-standard datasets that were able to be mapped onto the extracted sub-graphs. Clicking on a row in the list of mapped complexes highlights the overlap between PEPPER predicted complex and the considered mapped reference complex.

Known_mapped_complex

For each mapped complex onto a predicted sub-graph, sensitivity, precision and accuracy are computed to provides indicators of the goodness of the overlap between the two complexes. Matching-score is also computed. Results overall representation in terms of performances is available by selecting a specific complex in the Select... drop-down list.

  • Gene Ontology (GO) terms that are significant for the predicted complexes are displayed in the Ontology tab. Tables provide results for GO terms from the Biological Process (BP) and Cellular Component (CC) classes. Again, clicking on a specific term highlights which of the predicted complex proteins are associated with it.

Ontology

For each significant GO term, its systematic and common names are displayed alongside its q-value extracted from a single-sided Fisher's test and its number of occurences in the given predicted complex.

Other

Node details

Cytoscape Table panel Node Table includes post-processing characteristics which were computed for each of the proteins present in a network generated by PEPPER.

Table_panel

For each protein, its name and hypothetical weight attributed from a biological experiment are displayed alongside its bait status (isBait), whether it was removed from the complex by PEPPER algorithm (outcast) or if it is a protein added by the latter algorithmic procedure (isExpansion).

Topological gains while comparing the predicted complex without and with expansion proteins are displayed for the following computed features: degree (gain_degree), clustering coefficient (gain_clustCoef) and closeness centrality (gain_closeness). The number of reference complexes that were mapped onto a specific protein is also available as #complex whereas its number of annotations in GO terms is #annotations. The lists of corresponding significant GO BP and GO CC terms are also visualisable respectively in the GO_BP_terms and GO_CC_terms columns.

Scores for each of the post-processing characteristics are available too: ontology scores for GO BP and GO CC significant terms (GO_BP_score and GO_CC_score), topology (topology_score) and known complexes mapping (knownCplx_score). An overall score which is the average of the four previous ones is also available as postProcess_score.

Network details

Cytoscape Table panel Network Table provides an overview of all the GO terms that were predicted as significant for a given complex predicted by PEPPER.

Network_table

Algorithm parameters

Genetic algorithm parameters are customisable by clicking on the Advanced settings... button in PEPPER input panel or under the Apps -> PEPPER -> Options sub-menu item.

Algorithm_parameters

Crossover and Mutation parameters are evolution operators of the genetic algorithm expressed as probabilities. Iteration number corresponds to the number of steps of the algorithm whereas Population size is the size of the populations of the algorithm. Finally Solution size denotates the number of solutions retrieved from the Pareto front.

Contact

If you have any question or comments about PEPPER please contact us at pepper.plugin@gmail.com.

Location - contact

iSSB
CNRS UMR8030 - UEVE - CEA - GENOPOLE 

5 rue Henri Desbruères
Genopole Campus 1, Bât. 6
F-91030 ÉVRY CEDEX, FRANCE

contact (at) issb.genopole.fr

Phone: +33(0)1 69 47 44 30
Fax: +33(0)1 69 47 44 37

iSSBbuildingsmall.jpg

To visit us, please follow Genopole access indications.  Once on Genopole campus 1, iSSB is located on the first floor of Building 6, which is the first building to your left as you cross the pedestrian gate, or to your right as you cross the car gate.

GPS

Longitude: E 2° 27' 4''
Latitude: N 48° 37' 7''

 


Manip.jpg

 

qPCR.jpg

 

PCR.jpg