Why is the number of genes in certain a module in WGCNA super high?
0
0
Entering edit mode
5.3 years ago
Laura_zz • 0

In my WGCNA analysis, I got a total of more than 20 modules. Most of the modules ranged in numbers from a few hundred to several thousand, but one module had a gene count of 13,000. I don't know what is wrong with this. ? My softpower value corresponds to R^2 up to only 0.74. Is this the reason?

WGCNA • 3.4k views
ADD COMMENT
0
Entering edit mode

Is it the grey module, by any chance?

ADD REPLY
0
Entering edit mode

turquoise module. I used the block wise method to build the module, maxBlocwise set 20000.

ADD REPLY
1
Entering edit mode

You may gauge more by analysing the dendrogram of the genes, if you have generated that? With WGCNA, I find, one does usually end up with one large module.

ADD REPLY
0
Entering edit mode
I find, one does usually end up with one large module.

And is that usually significant to a trait? Even I have the turquoise module with almost half the genes in analysis and it is negatively correlated to a trait.

ADD REPLY
0
Entering edit mode

It is negatively correlated but is it statistically significantly correlated?

ADD REPLY
0
Entering edit mode

Yes. It is significantly negatively correlated. Cor = -0.82, p-value=9e-11.

ADD REPLY
0
Entering edit mode

Are you sure that they are not just genes that have no expression? HOw many genes are there? are they enriched for any pathway?

ADD REPLY
0
Entering edit mode

Turquoise module contained 10998 genes. A majority of the differentially expressed genes occurred in this cluster. Using PANTHER GO, I found several terms that were significant but I am doubtful as the terms seemed to be more general than specific.

Recap: I am performing this analysis on 40 samples from two conditions, normalized in DESeq2, batch correction by limma. Hierarchical clustering and sample-trait relation showed clear grouping of the sample based on condition. SoftThresholdpower=4 (at 0.8); Unsigned network; mergeCutHeight = 0.25; DeepSplit=2.

Images: SoftThreshold ClusterDendrogram

ADD REPLY
1
Entering edit mode

Not much to add, really, except that I would not permit that WGCNA's results dictate your conclusions for your work. It may make sense, given your experiment, to have roughly half of the protein coding transcriptome appearing in one module like that - I'm not sure.

ADD REPLY
0
Entering edit mode

To understand what is going on with your data in WGCNA, plot the eigengenes of each module (moduleEigengenes). If you do not have any outliers, there shoud be an agreement between the experimental design and the module eigengene

ADD REPLY

Login before adding your answer.

Traffic: 1693 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6