I am running Q2-2018.6. I imported featuredata[sequence] and featuretable[frequency] from USEARCH. That run contains multiple projects and I wish to split it by sample. Should I do this before I build the mafft alignment and phylogenetic trees, creating multiple trees for each project? Alternatively, I would build the tree with all of the samples in all projects and then filter using a subsetted metadata file for core metric or diversity analyses.