Source Tracker Use Question

[Roxanne Banker]

Hello, 

I am using Source Tracker to determine the potential source of microbes found in my samples (obviously). What I wanted to know is does Source Tracker have a function that can tell me which taxa are likely to have come from a certain source to a certain sink, or does it only predict the relative contributions of each source to each sink?

 If the latter is true, is there a QIIME script that will allow me to get at this data? I am looking through the QIIME scripts page, but I am still new to the program and I am not sure what might work best.

Best,

Roxanne

[Dan Knights]

Hi Roxanne,

The newest version of SourceTracker (0.9.7) by default outputs this information to a subfolder called full_results. This folder contains one tab-delimited table for each source environment (including Unknown). In each table the rows are the samples, the columns are the OTUs/taxa/microbes, and the entries are the predicted relative contribution to that sample of that microbe from the given source environment.

Dan

[Roxanne Banker]

Hi Dan,

Thank you for the help! I used the new version of Source Tracker to analyze my data and found the full_results folder that you mentioned. The .txt output lists all of the OTUs/taxa using a sequence ID. I wanted to know how I associate this number with taxonomic information. Do I need to add metadata to the OTU biom file before running Source Tracker or is there an additional script I need to run after the fact?

-Roxanne

[Dan Knights]

Hi Roxanne,

An easy way to link the OTU IDs to the taxonomy is to convert your OTU biom table to a classic tab-delimited text-based table with biom convert, and make sure that you extract the taxonomy. You will need to use the parameter 

"--header-key Taxonomy" or "--header-key taxonomy" when doing the conversion. Then open the table in Excel and the last column will tell you the taxonomy for each OTU.

An alternative is to rerun SourceTracker using a species- or genus-level taxonomy table instead of an OTU table, in which case the names of the features in the output tables will actually be taxonomic classifications.

Dan