Hi,
I construct my reference and validation network via WGCNA. In both network, I found similar modules in 2 network that one of them is green. Also, that is my prognostic module and I would like to find hub genes in the green module. My green module in Ref. Network has 948 genes and in validation network has 661. By intersect() function, I compare gene symbol in two modules but I found just 5 genes as common. My network type is “Signed” and the correlation method is “Pearson”. I don’t know why does it happen and which factors have effect in my study. That is important for me because I have found some hub genes in the green module in my ref. network and for validation that I need to find that hub genes into the green module in validation network. Generally, do I have another way for hub gene validation?
I appreciate if anybody share his/her comment with me.
Hi modarzi, what are you asking is not clear. Are you trying to compare two networks: Ref vs Valid, to see if the hubs in the ref are also hub in the Valid dataset?
For now I can tell you that the module color does not have anything to do with the gene content. In wgcna colors are assigned to modules according to their size: e.g. the largest module is the Turquoise, the second largest module is the blue, and then you have the red, brown, yellow, green, red and go on... In order to check if the genes in the green module of the Ref network still cluster together in the Valid network, you can do that by computing a contingecy table: see here. However, this approach does not give you any information about the 'preservation' of the hub across the two networks.
Ηι Andres, thanks for your comment. I know the module preservation process and I did it for my study. but my green module is a module that significant correlation with my trait. naturally, finding hub genes in this module is important.based on that I used the highest intra-module connectivity measure for finding hub genes. but I think these e.g top 50 genes should be validated. one way is finding these 50 genes in the green module at the validation network. am I wrong? do you have a better solution for the gene hub validation process? I appreciate if you share your comment with me?
I could be wrong but, supposing your network is scale free, or approximate to a scale free topology, and you already did the module preservation analysis, does the green module in the Ref. network results preserved in the validation network? If the answer is yes, then you can assume that the hub genes of the green module in the Ref. Net are also hub genes in the Valid. Net. In this case I would look at the stats Zconnectivity.pres; a value lower than 2 indicates that the hub is not preserved in the Valid. Net. Also, you said that the green module correlates with your trait of interest. Could you run again the module-trait relationship analysis with the Valid. dataset by imposing the module assignment of the Ref. network to check if the green module still correlates in a significant way with your trait of interest?
Thank for your comment. all my modules in the preservation process fall between 2-10 Zconnectivity.pres. It means my modules preserve in middle mode. not poor, not strong. unfortunately, My validation data set has no trait the same as ref. data set. In fact, my validation data set includes only the gene expression profiles whit out, same ref. trait data set. based on my condition I appreciate if you share your solution for validating hub genes in ref. Net.
At this point I would experimentally validate my hub genes. Perhaps only those genes that might have a role in the trait of interest that significantly correlate with your module. Check the literature.