WGNCA (correlation network): how can I incorporate mutant and wildtype into the same network analysis?

Question

[heyang]

Hello Everyone,

I have RNAseq data set which I looked at the expression profile of a wild-type and gene knockout mutant at two pHs (7.4 and 5.0). I want to identify modules that are associated with pH and genetic background. So for the trait files, I used binary values (0,1) to indicate whether it is that pH and genetic background, for example: pH7.4-WT.

The trait files looks like this:

Sample  pH5.0_WT    pH5.0_mut   pH7.4_WT    pH7.4_mut
pH5.0_WT_rep1   1   0   0   0
pH5.0_WT_rep2   1   0   0   0
pH5.0_WT_rep3   1   0   0   0
pH5.0_mt_rep1   0   1   0   0
pH5.0_mt_rep2   0   1   0   0
pH5.0_mt_rep3   0   1   0   0
pH7.4_WT_rep1   0   0   1   0
pH7.4_WT_rep2   0   0   1   0
pH7.4_WT_rep3   0   0   1   0
pH7.4_mt_rep1   0   0   0   1
pH7.4_mt_rep2   0   0   0   1
pH7.4_mt_rep3   0   0   0   1

By doing this, I got the module-trait graphs which has 4 columns corresponding to the 4 traits. Below:

Is this the correct way to do it? because I don't actually have a quantitative trait.

(Just learnt that the it is recommended to have more at 20 replicates (which I don't have at all), and I suppose the average of these replicates were used instead of like this).

Thank you!

[Ram]

Please use the formatting bar (especially the code option) to present your post better. You can use backticks for inline code (`text` becomes text), or select a chunk of text and use the highlighted button to format it as a code block. I've done it for you this time.

Also, tabular data can be displayed much more elegantly when the column command is used well.

WGNCA

Surely you mean WGCNA? Shouldn't that be a tag so the question pops up on a tag based search?

[heyang]

Thanks for the advice, did change it.

Answer 1

[Kevin Blighe]

Well, at least you have different statistically significant correlations for each group, but what do they mean (?) - take a look at the genes behind each statistically significant module and see if you can infer anything from them.

Keep in mind the following:

1. there are many different ways that you can approach your analysis, given the data that you have
2. WGCNA will not conclusively give you a result for your study - at best, it may simply guide you. Be conscious of 'cherry picking' results that you want to find

Kevin

[1769mkc]

"such as regression modeling" you had one post which showed regression modelling in biostar , can you provide the link ,i did look but couldn;t find it, as i couldn;t recollect the question you had answered for ..

[Kevin Blighe]

Hey, do you mean either of these:

• A: What is the best way to combine machine learning algorithms for feature selectio
• A: Resources for gene signature creation