Entering edit mode
5.9 years ago
hennet.lauriane
▴
40
Hello,
I ran a coexpression analysis with WGCNA on my full transcriptome data and I am trying to understand what genes are likely to be important for the process I am interested in. I use a set of genes that I am sure they are involved in this process. I chose to extract genes lists co expressed with these genes based on their coexpression coefficient. I use a "high" threshold and run enrichment analysis.
Is it a usable approach on its own?
I ask this question because I also see other WGCNA metrics used in paper such as links between modules, links with phenotype (data that I don't have) etc.
Thank you
In WGCNA, as per the tutorial I dont need the modules results in graph . Can anyone suggest me how to extract lists/tables or modules summary (results)? If possible kindly help we with the lines of code too.
..but how can we help you? You have not fully described any type of data that you currently have, nor have you described how it was produced (?).
Here's a tip to help you, though: make use of the
str()function in order to see the internal structure of objects. In this way, you'll be able to deduce which object internal components you need to access for writing out to disk.WGCNA will never completely shine light on any disease mechanism. I would not use it as a priority in any study. That said, if you want help, can you please share the code that you have used? For example, when you say that you "extract[ed] genes lists co expressed with these genes based on their coexpression coefficient", how did you do this in the context of WGCNA? Outside of the context of WGCNA, your statement just implies that you took the genes that have high positive correlation to the genes that you know are involved; so, it's just a correlation analysis.
Hey, sorry never saw this answer and forgot about my post. Thank you for the help. I don't use a code on its own, I actually export all highly coexpressed genes in their corresponding module (as if I wanted to use them for Cytoscape vizualisation) and I extract gene that are coregulated with my genes of interests.
My lists/tables are as follow;
moduleblue gene1 gene6 weight1
moduleblue gene2 gene5 weight2
moduleblue gene3 gene4 weight3 . . .
If gene3 is of interest for me I will go get every gene4 that is linked to it (only considering a weight above my chosen threshold).
Indeed, it is just a correlation analysis, but is WGCNA anything else but this ? Thank you
It is indeed more than a simple correlation analysis. Please read through the website to find out what it does: https://horvath.genetics.ucla.edu/html/CoexpressionNetwork/Rpackages/WGCNA/Tutorials/
What you could now do is perform gene enrichment on the genes in each module. You can use topGO and KEGGprofile packages
Yes indeed WGCNA package is more than this but what I understand is that the direct output is a set of correlation lists. I stoped using WGNA as this stage as I don't have external traits to input and already started on enrichment analysis as you suggested.
Thank you very much Kevin
You are welcome.