q2-evident: repeated measures

Hi developers,

Thank you for the plugin. I hope this is the right way to ask.

So, I tried to install it to the current qiime2 conda environment and it returned an error message: PEP 517 building error. I then installed pip inside it by conda install pip, and it solved the problem. I wonder if it is a correct fix though.

Then, we wanted to analyze our samples with the
qiime evident univariate-power-analysis-repeated-measures.

We used shannon and numeric type of age column
--p-state-column age.

It then returned the following error:

qiime evident univariate-power-analysis-repeated-measures \
> --m-sample-metadata-file metadata.tsv \
> --m-sample-metadata-file shannon.qza \
> --p-data-column shannon_entropy \
> --p-individual-id-column subject_trans \
> --p-state-column age \
> --p-subjects 4 8 \
> --p-measurements 2 \
> --p-alpha 0.05 \
> --p-correlation -0.5 0 0.5 \
> --p-epsilon 0.1 \
> --o-power-analysis-results power.qza

Plugin error from evident:


Debug info has been saved to /tmp/qiime2-q2cli-err-hf9x6x0a.log

We then edited the metadata to change the data type from numeric to categorical and run a similar command, the error changed to this

Plugin error from evident:

  Cannot calculate effect size of repeated measures with missing values.

Debug info has been saved to /tmp/qiime2-q2cli-err-na1rxgsu.log

So, what does this error mean? I assumed it has something to do with
> --p-subjects 3 8 \?

Thank you very much!

1 Like

Hi @Diki

At the moment Evident only supports calculations on categorical groupings. If age is a numeric column then unfortunately the repeated measures analysis won't work here. We do hope to add numeric column support in the future, however.