Hi, Iwould like to ask for your advice on whether WGCNA is appropriate for my paired microarray dataset, and how I should proceed given the very small sample size and multiple interventions.

My study contains 12 samples from 6 patients. Each patient undergoes two cycles, which allows us to account for intra-individual variation. There are two intervention cohorts: three patient pairs are exposed to hormone versus their corresponding controls, while the other three patient pairs are exposed to compound X versus control. Importantly, the control samples are the same type across all six pairs. My primary research question is whether hormone exposure induces transcriptomic changes compared with the control condition.

Because of the small sample size, I initially applied removeBatchEffect from the limma package to account for the paired/inter-individual effects while retaining all 12 samples in the analysis. Using WGCNA, I cannot identify any biologically meaningfull results. I know my data is too small to reach scalefree topology cuz my picture for softthreshold detection that was weird. The power is up and down under 0. My question is whether I should construct wgcna using a larger and public dataset with having same study design in human samples and same question in transcriptomic changes but different intervention then do module concensus/preservation to my dataset. I means use another data as a reference data and my data is a test one. Hope to have response from you . Thank you!!!!