I will be grateful for your feedback and advice. I have been tenured for some time now and appreciate that my statistics training was very different than what is available now.

Can I use WGCNA to identify clusters of responsive proteins from one perturbation time series (pooled from two independent replicates)? I plan on following up with additional proteomic experiments, based on specific clusters that I expect will address a priori hypotheses, to verify DE between pre- and post- shift conditions.

Specifically, I am working with a microbial population subjected to a shift from iron limited to iron replete conditions. I have determined the time scales when cells recover from iron limitation (a photosynthetic parameter) and when they sequester the most stored iron (radioisotope-based). I am interested in collecting samples for proteomics during two identical time shifts, with sampling times informed from these ‘phenotypes’. Since I am using TMT16 isobaric labels in the shift experiment, I am conducting two sets of independent time series samples (to insure against or account for the, “what if there was an unknown effect that had nothing to do with your treatment” phantom) and pooling them. Note that this will not include a control time series where iron is not added. For many reasons this would not really be a control (don’t want to bog you down with the details).

Does this seem reasonable?

And, since I am not testing for DE in gene expression profile compared to a control time series so I can use the sample space to best resolve the temporal responses, I am considering staggering the independent samples. This would make, for example, sample 1 = times 0, 1, 3, 5, 7, 15, 17, 19 h and sample 2= times 0, 2, 4, 6, 8, 16, 18, 20 h. (these two time windows capture the two changes in phenotype).

Thank you!

Adam ```