= Cytoscape Workflows =

This page describes different ways that Cytoscape is used by researchers.  The intent is to capture different workflows to ensure that we support the different steps in each process as well as the transitions between steps.

The following workflow is roughly what is described in that Cytoscape ''Nature Protocols'' paper that will be published in the fall of 2007.

 1. '''Load networks.'''
  * Agilent LitSearch
  * cPath
  * Pathway Commons?
  * IntAct and/or DIP?
 1. '''Load attribute, synonym, and annotation information about the networks.'''
  * GO annotations
  * attribute files
  * synonyms?
  * Gene expression data (ArrayExpress?)
 1. '''Map attributes to networks'''
  * BioMart ID mapping.
  * [[WebServicesIDMapping]]
 1. '''Analyze and Visualize the networks.'''
  * Apply layouts
  * Apply VizMapper
  * MCODE
  * jActiveModules
  * BiNGO
  * Cytoscape's built in tools like filters, selection, etc..
 1. '''Make joural quality images and publication materials.'''
  * Export graphics
  * Export session to web

== Alternative Workflow 1 ==

This is the most common workflow Gary has seen for biologists studying interactions

'''Before loading data into Cytoscape'''
 1. Collect interactions experimentally 
 1. Evaluate false positive, false negative rate
    This is not currently done in Cytoscape, although it would be nice if it could be. It can be done by comparing the user interaction set against a set of known high quality interactions - something like the merge plugin, but it would need identifier mapping to work well + some simple overlap stats reported.

'''After loading data into Cytoscape'''
 1. View network -> publication quality figure (yFile organic layout)
 1. Zoom into genes of interest -> publication quality figure (no current layout does a good job with zoom in diagrams)
 1. Optional: predict gene function based on new interactions

== Alternative Workflow 2 ==

The most common workflow for Gary is:

'''Before loading data into Cytoscape'''
 1. Predict interactions computationally
 1. Evaluate false positive, false negative rate
    This is not currently done in Cytoscape, although it would be nice if it could be. It can be done by comparing the user interaction set against a set of known high quality interactions - something like the merge plugin, but it would need identifier mapping to work well + some simple overlap stats reported.

'''After loading data into Cytoscape'''
 1. View network -> publication quality figure (yFile organic layout)
 1. Zoom into genes of interest -> publication quality figure (no current layout does a good job with zoom in diagrams)
 1. Optional: predict gene function based on new interactions