Issue generating regression_summary.qzv with q2 gneiss differential abundance


I am currently using the Differential Abundance Analysis with gneiss, and am having issues generating the regression_summary.qzv.
I was able to generate the hierarchy.qza and balances.qza, but have issues when I get to the regression step.
It seems to be an issue with my metadata file. As I keep getting this error:

Plugin warning from gneiss:

ols-regression is deprecated and will be removed in a future version of this pl$
warnings.warn(‘Cannot determine yajl version, assuming <1.0.12’)
Plugin error from gneiss:

Error evaluating factor: NameError: name ‘sampleID’ is not defined
I have been searching on all forum posts and it seems others had this issue because of blank spaces in their metadata file, not having sampleid column, and the metadata file not being a tsv file. I have addressed these issues and I am not sure what is wrong at this point. Any insights?

I am using qiime2 version 2019.10 installed with conda.

This is the code that I ran:

qiime gneiss ols-regression
–p-formula “sampleID+year+colarea+date+ageclass+massgained”
–i-table balances.qza
–i-tree hierarchy.qza
–m-metadata-file sample-metadata2.tsv
–o-visualization regression_summary.qzv

I verified my metadata file with Keemei and it checked out. Let me know if you would like me to attach that.

Thank you for your time!


Welcome to the forum @ginajohnson!

“sampleID” is probably the name of the sample index, so it will not actually be considered a metadata column that you can use in analysis. You should remove that value from your formula and all should be well… besides, presumably sampleID is unique for each sample, in which case it would not be usable in the formula.

But perhaps I misunderstand — if I have, please share your metadata file so that we can take a look.



Thank you for the reply. I did this and now and getting this error back:

Plugin error from gneiss:

cannot convert float NaN to integer

I have attached my metadata file.

Thank you!
sample-metadata2.txt (1004 Bytes)

Hi @ginajohnson,

This is due to missing values in your metadata. Looks like you have one sample (“M–”) that has all NA values, maybe a blank? Remove that from your feature table and then give that a spin!

This fixed my problem! Thank you!!

1 Like