Iggy-POC

Iggy-Poc is a graph coloring model created by Bertrand Miannay to :

• Identify subnetworks from a prior knowledge network (PKN).
• Compute the maximal similarity with each subnetwork from a set of observations.

This model need Python 2.7.6, the package NetworkX and Clingo 4.5.4 ( https://sourceforge.net/projects/potassco/files/clingo/4.2.1/)

To try a toy example, the toyExample.sh file allows to use the framework on a small network :
chmod 777 toyExample.sh
./toyExample.sh

Files :

• graphCompaction.py

  • Example : python tools/graphCompaction.py graphToyExample.sif ReducedGraph DicoNodes grapheMEF.lp

  • Generate a reduced graph based on subcomponents identification

  • Input :

    • graphToyExample.sif : PKN file respecting sif format (A 1 B)

  • Output :

    • ReducedGraph.sif : Reduced graph respecting sif format
    • DicoNodes : Hash table ("A +" : node1)
    • grapheMEF.lp : reduced graph respecting lp format (edge(node1,node2,1,2).)

• optimizationComponent.lp

  • Example : clingo grapheMEF.lp tools/optimizationComponent.lp -n 0 --opt-mode=optN --enum-mode=cautious --quiet=1 | grep "correle"| sort | uniq | sed s/" "/"\n"/g | sed s/")"/")."/g > correlations.csv
  • Identification of correlated nodes in the graph based on the perfect colorations constraints
  • Input :
    • grapheMEF.lp : reduced graph respecting lp format (edge(node1,node2,1,2).)
  • Output :
    • correlations.csv : correlation between nodes (correlePositif(node2,node1).)

• componentIdentification.py

  • Example : python tools/componentIdentification.py DicoNodes correlations.csv > temporyFile
  • Input
    • DicoNodes : Hash table ("A +" : node1)
    • correlations.csv : correlation between nodes (correlePositif(node2,node1).)
  • Output
    • temporyFile : Unsorted components ("B +, C -, A +")

• componentsSort.py

  • Example : python tools/componentsSort.py temporyFile | sort > components.csv
  • Input
    • temporyFile : Unsorted components ("B +, C -, A -")
  • Output
    • components.csv : Sorted components ("A +, B -, C +")

• MSComputing.py

  • Example : python tools/MSComputing.py dataExample.NA components.csv >> resultat_MS.csv
  • Input
    • dataExample.NA : Observations (A = 0)
    • components.csv : Sorted components ("A +, B -, C +")
  • Output
    • resultat_MS.csv : Maximal similarity for each component