That sounds right, except that you may want to include assigning taxonomy as well . The outputs from Qiita do not automatically have taxonomy associated, but running the merged reference-hit.seqs.fa data through qiime feature-classifier classify-sklearn should do the trick.
Great! last question - Does Qiita automatically merge and provide the reference-hits.seqs.fa when I merge deblurred files or do I do this myself ?Thanks!
reference-hits.seqs.fa is not provided. However, this file just contains the different sequences in my biom table right? So I could just generate a new file that has all of my sequences beginning with "> " and use this?
To summarize for others interested in SourceTracker, I
downloaded Deblur output files called reference-hit.biom from each of the studies I was interested in to my computer, instead of merging them all via Qiita and downloading. I think the file from Qiita with all the data merged was too big to for my computer to process the file
converted the biom files to txt format and took a subset of samples I was interested in (there were thousand samples in some of the studies) and then converted these back to biom format. This step was because I didn’t know how to edit biom files
$ biom convert -i HumanSkinSamples.txt -o HumanSkinSamples.biom --table-type=“OTU table” --to-hdf5
merged all the biom tables together using qiime1 merge_otu_tables.py
$ merge_otu_tables -i HumanSkinSamples.biom,HumanFecalSamples.biom -o merged.biom
imported merged biom table as a qiime artifact
$ qiime tools import --input-path merged.biom --type ‘FeatureTable[Frequency]’ --source-format BIOMV210Format --output-path merged-feature-table.qza
created a reference-hit.biom.fa file using excel tricks (saw your code after I already did this) and imported as a qiime artifact
$ qiime tools import --input-path rep-seqs-merged.fa --output-path rep-seqs-merged.qza --type FeatureData[Sequence]