Do we need to have identical samples for recorded traits and expresseion data in order to calculate the correlation between traits and modules?
What do you mean?
I mean that do we need to have trait data for all samples that we have RNA-seq data for?
It would help. Why would you have any sample for which there was no trait / metadata?
Data that we have are from separated experiments, therefore for some of the samples we do have expression data, but no trait data