Using the SVM FeatureFinderIdentification workflow with different biological samples


I’ve been running successfully the original FeatureFinderIdentification (FFID) workflow (from Hendrick Weisser’s paper) using the machine learning (SVM) option. I have 2 different biological types of samples that I want to compare: triplicates of nuclear and cytosolic fractions extracted from the same cells.

To remove potential biases, I’m wondering if I should analyze separately each type of samples through FFID by comparing replicates only, like nucleus replicate #1 as internal ID and nucleus replicates #2 & #3 as external IDs. Or is it okay to perform the analysis using all the samples (nucleus & cytosol) as external IDs, regardless of the biological types of samples?

Many thanks!