can I use fold change instead of expression value for coexpression network construction analysis? We have performed global proteomic analysis between 3 conditions: (1) control (2) disease A, and (3) disease B. Objective is to check the differences in the PPI.
(Note: we have same set proteins between both the disease only difference is fold change)
I don't think it's appropriate to use fold change for co-expression network construction. Simple reason is fold change's a pooled value and for correlation anything less then 3 samples would be a problem. Infact WGCNA requires minimum of 15 sample size. I would suggest you to use individual samples
@maddy thank you for the quick response. We have performed the analysis on triplicates of two samples. Other issue is we have only control to case ratio values. Could you please suggest what else can be done??
With you sample size, it is rather unrealistic to build co-expression network and it might be better for you to find known networks that were enriched in your samples.
If you have the p-value of the difference, you can use the download a ppi information and look for enriched subnetworks using Cytoscape. You can look into this but I am certain there is one function which helps you to reduce the network down to the significant subset and it is based on this paper: Discovering regulatory and signalling circuits in molecular interaction networks, maybe you can try and find the function yourself as it was 2 years ago when I used it.
ADD COMMENT
• link
updated 2.3 years ago by
Ram
44k
•
written 9.0 years ago by
Sam
★
4.8k
@sam can you repost it please? The link shows not found. I have a relatively large data size and I am also trying to createe a network only based on f.c values.
@maddy thank you for the quick response. We have performed the analysis on triplicates of two samples. Other issue is we have only control to case ratio values. Could you please suggest what else can be done??