WGCNA TOMType specification
2
1
Entering edit mode
@aurorasavino-13948
Last seen 6.6 years ago

Hello,

I have a question regarding the TOMType parameter in WGCNA blockWiseModule function.

I am interested in distinguishing positively and negatively correlated groups of genes, therefore I set networkType = "signed". By mistake, though, I put TOMType = "unsigned". Specifically:

net = blockwiseModules(data ,power= 6, TOMType ="unsigned", corType="pearson", networkType = "signed", minModuleSize = 30, reassignThreshold = 0, mergeCutHeight = 0.25, numericLabels = TRUE, pamRespectsDendro = FALSE,saveTOMs = T, verbose = 3)

I looked at the documentation but it still isn't clear to me what choosing TOMType = "unsigned" implies.
Additionally, I ran the function again changing to TOMType="signed" and the resulting modules seem to be very similar (almost all the same genes).

Is choosing NetworkType="signed" enough to retain the sign in the resulting modules?

Many thanks in advance for your help.

Aurora

 

WGCNA network analysis Sign • 4.8k views
ADD COMMENT
1
Entering edit mode
@lluis-revilla-sancho
Last seen 8 days ago
European Union

TOMType options handles how should conflicting TOMs be handled (TOM between gene A and B is +, between A and C is +, and between B and C is -), usually there are few conflicts that's why it doesn't make much of a difference.

The NetworkType option is described in the adjacency function. There it says:

for type = "unsigned", adjacency = |cor|^power; for type = "signed", adjacency = (0.5 * (1+cor) )^power; for type = "signed hybrid", adjacency = cor^power if cor>0 and 0 otherwise

So if you want to keep the genes positevely correlated in the same module you should use the "signed hybrid" option.

ADD COMMENT
0
Entering edit mode
@aurorasavino-13948
Last seen 6.6 years ago

Thank you very much Lluìs, sorry I hadn't seen your reply.

Would you suggest using "signed hybrid" in general? Is there any advantage except being sure to keep only positive correlations?
If I look at genes belonging to my modules they seem to be positively correlated, but indeed from the specifications it could be that I included some negatively correlated ones.

Thank you again!

Aurora

ADD COMMENT
0
Entering edit mode

That would depend on what you are looking for, if you want to have genes that follow the same pattern then use signed hybrid, otherwise use unsigned, or if you want to be in the middle use the "signed" option

ADD REPLY

Login before adding your answer.

Traffic: 975 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