I am using QIIME2-2017.11, and my code is:
qiime feature-table summarize
–i-table table2-dn-97.qza
–o-visualization table2-dn-97.qzv
–m-sample-metadata-file …/Data/Metadata/Metadata2.tsv
The error message is:
Non-string Metadata index values detected. There may be more errors present in this metadata. Sample/feature metadata files can be validated using Keemei: http://keemei.qiime.org
Just a couple of guesses that may be worth trying out while you wait for an expert answer on this.
I noticed your link to the google-sheet is titled 'Metadata1', while the file you input in to your code is 'Metadata2.tsv'. Are you sure you are putting forth the correct validated metadata file into qiime?
Are you importing the file from google-docs directly as a .tsv or in some other format and then manually converting it to .tsv in a second step? It might be that somehow in that second conversion you're losing the appropriate formatting.
A few of your column-headers have brackets ( ) in them, though it doesn't explicitly say if brackets are considered "certain special characters" in the metadata requirement, if they are, this could be problematic. My guess is since Keemei didn't identify them problematic, this isn't the issue, but it may be worth a desperate try?
This one is a bit out there (read: desperate) but I think has potential . You have 2 column labels that start with a # sign, and while the number sign is allowed anywhere according to metadata requirements,
"Comment lines (i.e. lines that begin with #) may appear anywhere in the file and are ignored."
I think in this case it may lead to violating another requirement:
The column labels must be unique (i.e. no duplicate values) and cannot contain certain special characters (e.g. /, , *, ?, etc.).
I'm not exactly sure how the number sign is handled here but if it leads to ignoring all the proceeding characters as I'm guessing it does, this would lead to your file having 2 empty field as column labels, which would violate the "must be unique" requirement, or maybe that empty headers are just not accepted in general.
Many thanks! You are brilliant! I got a feeling that the point #4 you mentioned would probably be the crux of the problem, so I checked it first! As expected, it worked!
Thank you again, and I guess this issue could be added to the metadata requirement!
Quick followup: the issue you ran into with # characters in your column names will be fixed in the upcoming 2018.2 release (a couple of other users have run into the same problem). We’ll follow up here when the fix is available in a release!
In the QIIME 2 2018.2 release, Metadata now supports column names containing # characters.
There are a number of other changes to QIIME 2 Metadata in the 2018.2 release. See this forum announcement for details on what changed, as well as the updated Metadata tutorial.
. You have 2 column labels that start with a 