Coverage and consensus subnetwork

[Subanshi Kumari]

Hi

I have generated subnetworks from hotnet2 (1.2.1) using only one protein-protein interaction network, but I have two queries

1. For all the generated subnetworks, hotnet2 has given the coverage value of 0.0%, (although the genes of these subnetworks are mutated in patient's sample)., how the hotnet2 is calculating coverage value?

2. As the hotnet2 also produces the consensus subnetworks from different PPI networks (multinet, HINT, Iref) (correct me if I am wrong), but as per my case, I am considering only one PPI network, then how the hotnet2 is predicting consensus output when I am running on single PPI network. Am I lacking in something?

Regards

[Subanshi Kumari]

Hi Matt

Can you please address the above queries?

[matthe...@brown.edu]

Hi Subanshi,

1. HotNet2 computes coverage of a subnetwork as the fraction of samples with at least one mutation in at least one gene in in the subnetwork.  However, depending on how you construct the heat scores, this information may not be available to HotNet2, so it may report 0.0% coverage.  For example, if you provide mutation frequency heat scores as input to the makeHeatFile.py script, then HotNet2 will not know which samples have mutations in which genes.

2. Given one network and set of heat scores, HotNet2 does not compute a consensus.  In this case, the HotNet2 "consensus" output is the collection of HotNet2 subnetworks corresponding to the minimum value of delta with the largest number of statistically significant subnetwork sizes (please see Section 1.4.2 of the supplement to the HotNet2 paper).

Best,

Matt

[Subanshi Kumari]

Ya 

But I have used mutation data (snv file) to generate heat score using makeheatfile.py.