Finding clusters of interacting proteins in a protein-protein interaction graph

Question

[rubic]

Hi,

Given a protein-protein-interaction graph (e.g., from the string data base) and an outcome of a differential expression analysis (i.e., for each gene we have its log fold-change and the associated p-value between the treatment and control), is there a graph-based algorithm that can find subgraph parts enriched with genes which log fold-change p-value is below a given cutoff?

Answer 1

[Diwan]

Given a list of differentially expressed genes, 'keypathwayminer' in cytoscape can extract the highly connected sub-network from PPI. I had used their older version and found it helpful.

Check here for app: http://apps.cytoscape.org/apps/keypathwayminer

Their recent paper: http://bmcsystbiol.biomedcentral.com/articles/10.1186/s12918-014-0099-x

To further group the sub-network, you can try 'clusterviz' app in cytoscape to get network clusters.

An application paper that uses both keypathwayminer and clusterviz: http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0112193