I have a dataset for proteomics, miRNA, mRNA profiling for three disease groups and controls. I want to see the similarities and differences between networks.
Few suggestions I would need before proceeding-
Do I do each dataset separately or out them all together and analyze?
Do I run all the phenotypes together?
For miRNA- do I do the negatively correlated pairs before adding?
Also, if anyone is familiar with the wgcna tutorial. I was trying to see the module association with clinical traits (here I have types, age, group, subject rather than any numerical measurement unlike the tutorial) Was wondering If i can still see the module correlation to these traits. In my current analysis, it just gives me NAs . I am not sure why.
1) Each dataset (analysis type) can be run separately and you can look for module conservation or determine some other correlation/connection between the modules for the individual datasets later. It wouldn't make sense to run them together because the data/units aren't on the same scale and the identifiers are likely different for each.
2) If the phenotypes have anything to do with each other (derived from the same tissue, or from the same patients, etc) then it makes sense to run them together. If there's no logical relationship between the phenotypes (if comparing phenotypes wouldn't be something you're interested in) then they should be run separately.
3) By this do you mean filter the miRNA for ones that are significantly regulated in a direction that is opposite of their targets? It might make sense. I'd probably try it both ways.
Also, if anyone is familiar with the wgcna tutorial. I was trying to see the module association with clinical traits (here I have types, age, group, subject rather than any numerical measurement unlike the tutorial) Was wondering If i can still see the module correlation to these traits. In my current analysis, it just gives me NAs . I am not sure why.
Thanks,
Mamta