HotNet --consensus_permutations

[James Roberts]

Hi there,

I have a question about the consensus permutations flag on HotNet2.py.   I see the default value is set to 0, is there a recommended starting value for getting p-values for the consensus results?

Thanks for your help.
BW,
James

[matthe...@brown.edu]

Hi James,

I apologize for the delayed response.  HotNet2 assesses statistical significance of the HotNet2 subnetworks for each combination of network and heat scores, and the consensus procedure combines HotNet2 subnetworks from multiple networks and/or heat scores.  As you noticed, HotNet2 does not assesses statistical significance for the consensus results by default; this is an optional, additional step.

If you only have one network and/or set of heat scores, or if you only need to evaluate statistical significance on each combination of network and heat scores, then I would choose zero consensus permutations.  Otherwise, I would choose as many as you need, but since the consensus procedure generates consensus results for each consensus permutation, 100 consensus permutations, for example, would take roughly 100 times longer than no consensus permutations.

Best,

Matt