Export data for use in Sourcetracker

Hi all,
With the switch from QIIME 1 to QIIME 2, I am looking to update my workflow. Previously, I followed a similar workflow as per the tutorial in the QIIME 1 documentation.

I have run basic analysis (see my notes) and reached a road block. I am unsure what is the best way to export frequency of OTUs into a tab delimited file for use in Sourcetracker. In QIIME 1, export of the OTU table would give me approximately 70,000 OTUs. If I export data from QIIME 2 via the taxa collapse, the table produced is approximately 1200 OTUs. Does anyone have suggestions on the best way to export data from QIIME 2 for use in Sourcetracker?

Hi @Chris_Main,

Coincidentally I’ve been playing with SourceTracker2 and it would be very easy to turn into a plugin…

If you are using an ASV method like DADA2 or Deblur, then you would expect to see far fewer OTUs. Depending on your goal, using the collapsed taxa, or using the ASVs themselves should be fine.

Looking at your notes, I think you’ve got the right idea. Are you running into mechanical issues getting everything pieced together, or is this more of a theory question?

Hi Evan,
My question stems from a few things. I am working on a pilot study for fecal source tracking in the waters of Delaware. Currently, we are focusing on a small watershed as a proof of concept but with aims to work throughout the state to examine impaired waters for potential future remediation. As such, I’m trying to write SOPs that can be done by others than myself if we decide to continue on this path. My question is what is the best way to get data from ASV to Sourcetracker/Sourcetracker2? Is the taxa collapse correct or is there a better option?

Hey Chris,

I would stick with just using straight ASVs. So what you have in your notes (simply using qiime tools export) will work great.

I do think you are supposed to rarefy your table first however (you can use the output of core-metrics-* or feature-table rarefy for that).

Hi Evan,
Thanks for the information. I’m going to play around with it before finding the best workflow for us. One thing I need to explore is why field triplicates aren’t matching up as closely with each other.

This topic was automatically closed 31 days after the last reply. New replies are no longer allowed.