Data sets with different conditions

[avkitex]

I'm comparing samples under different conditions from the same GEO/ArrayExpress data sets.

What is the best way:

1. Calculate iterativeWGCNA on full data set
2. Compare eigengenes values
   Or
3. Cut data sets to small ones (per condition)
4. Calculate iterativeWGCNA
5. Compare modules by overlapping genes in modules

[fossilfriend]

I can't definitively choose one way over the other because the answer to this depends on both the experimental design (including technical biases such as platform, normalization, etc) and the biological question you are exploring. In general, though -- if your goal is to look for similarities across conditions but your experimental design varies from condition to condition (e.g., different platforms) the consensus approach might be best. If you wish to explore a continuity of process (e.g., differentiation) or patterns of expression that segregate related conditions (e.g., tissue-specific expression) and can account/control for variation due to experiment bias, then you would work with the full dataset.

[avkitex]

For example I have normal and cancer samples in one data set.
And I want to find pathways differences between cancer and normal

I have about 10 data sets on different platforms from different vendors (including RNAseq).
So what would be the best way in your opinion?