Question: WGCNA and VisAnt
gravatar for ldyer2006
3.3 years ago by
ldyer200640 wrote:

Hello folks, I'm hoping that you can help me understand some of the results I'm getting from WGCNA when it comes to visualising a network.

I have managed to extract my modules from the data set, and graphed them against their trait data, and now I'm looking to produce a visualisation of the gene network for each cluster, which was my original aim in using WGCNA. Unfortunately, I am having some problems with every network looking completely identical and symmetrical.

When I open my network file in VisAnt, I am always greeted by a perfect circle, with a line of equal strength connecting every node in the network. When I investigate the text output itself, it seems fine, example below, but there is absolutely minuscule variation in the connection strength, with almost every value being 0.40 to 0.42...

Hibch Ndufb3 0 M0039 0.402953581463721
Hibch Acadl 0 M0039 0.413293496510016
Hibch Sdhc 0 M0039 0.402266127071595
Hibch Ndufs2 0 M0039 0.412983888989828
Hibch Iars2 0 M0039 0.410319082165815
Hibch Atp5c1 0 M0039 0.399168115779561
Hibch Nfs1 0 M0039 0.401863940513408
Hibch Ndufb5 0 M0039 0.402559122972561
Hibch Etfdh 0 M0039 0.411821948607482
Hibch Gatb 0 M0039 0.405798431101945
Hibch Atp5f1 0 M0039 0.402543447219872
Hibch Hadh 0 M0039 0.410937281224753
Hibch Cpt2 0 M0039 0.411218531292222
Hibch Atp5k 0 M0039 0.408706798256598
Hibch Mdh2 0 M0039 0.410767514376832
Hibch Adipor2 0 M0039 0.40568057005744

The list goes on like this, with the gene on the left hand cycling through my "top" hub genes, but the final value never exceeding 0.43 and never appearing lower than 0.38. I have even run through the export function with every single gene in the cluster (Just under 2000), and no single connection has a value outside of this range.

Is there some easy problem that I'm missing? I am simply using the "exportNetworkToVisAnt()" function as provided by the WGCNA package. Is the problem that none of my modules actually exhibit a single gene network?

Any help is greatly appreciated. Thanks in advance.

rna-seq • 2.0k views
ADD COMMENTlink modified 3.3 years ago by Jake Warner810 • written 3.3 years ago by ldyer200640

What was the edge weight cutoff you chose to export ?

ADD REPLYlink written 3.3 years ago by pixie@bioinfo1.4k

It was low, just around 0.02, to ensure that I was seeing all of the connections as I just chose 25 hub genes.

ADD REPLYlink written 3.3 years ago by ldyer200640

Is there any specific reason to choose 0.02 as threshold.

How to extract the network of only hub genes?

ADD REPLYlink written 11 months ago by Arindam Ghosh340
gravatar for Jake Warner
3.3 years ago by
Jake Warner810
Jake Warner810 wrote:

Your data look ok and the default setting on VisAnt is the circle you describe. Try clicking 'Layout -> Relaxing' to see if this gives you the desired result. It's also easier to click the nodes if you turn off fine arts: 'Options -> Fine arts'. I think there's a way to map TO with edge thickness but you'd have to look that one up.

ADD COMMENTlink written 3.3 years ago by Jake Warner810

Thanks, Jacob. That is somewhat re-assuring.

Selecting the layout -> relaxing option seems to generate a plot that is slightly more like what I was expecting, thank you.

But is it really acceptable that all of my gene network measures fall within such a tight range? VisAnt allows me to set weight thresholds anywhere between 0 and 1, doesn't the fact that my edge-weight range is so distinct from this imply that my analysis has gone wrong somewhere?

ADD REPLYlink modified 3.3 years ago • written 3.3 years ago by ldyer200640

I think it's to be expected that all the values are in a tight range. I'm assuming though you exported the top 500-1000 connections which would include several hub genes that are all highly connected but maybe with relatively equivalent connection strengths. When you say you looked at all 2000 genes in the module, does that mean all 2000000 connections?

ADD REPLYlink written 3.3 years ago by Jake Warner810

Yes, when you put it like that, it makes much more sense. I have since re-run the analysis and now that I've looked more into it after these posts, I think I have a slightly better handle on it and the networks are at least beginning to look like networks. I think, with a lot more fiddling and a lot more reading of how to analyse this particular visualisation, I should be able to get something beneficial.

Thanks for your time.

ADD REPLYlink written 3.3 years ago by ldyer200640
Please log in to add an answer.


Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.3.0
Traffic: 1967 users visited in the last hour