qiime2-amplicon-2025.7 module 'skbio.diversity.alpha' has no attribute 'sobs' error

[ankurnaqib]

Hi. I ran this command-

qiime diversity core-metrics --i-table dada2-ccs_table_filtered_v1.qza --p-sampling-depth 49000 --m-metadata-file metadata_QIIME.tsv --output-dir core_metrics_49000 --p-n-jobs 8 --verbose &> core_metrics.log

and here is the error I get

AttributeError: module 'skbio.diversity.alpha' has no attribute 'sobs'
Plugin error from diversity:
module 'skbio.diversity.alpha' has no attribute 'sobs'

See above for debug info.

I even tried the fix mentioned on this link #566 (comment) but that also didnt help. I am attaching the log file as well. Please help. Thanks.

core_metrics.log

[fedarko]

Strange; thank you for reporting this. I was able to reproduce the issue on my system (and have enclosed steps at the end of this comment for reference).

Cause of the issue

For some reason, installing Empress causes scikit-bio and scipy to be downgraded (from 0.6.2 → 0.5.9 and from 1.13.0 → 1.10.1, respectively). So, this is likely related to #566 somehow.

It's very strange, because Empress' requirements (at least for v1.2.0, the latest version on PyPI) are all very lenient; it barely pins anything.

My guess is that this is related to the Cython / IOW / scikit-bio stuff discussed in #566. I noticed that running just pip install iow in the QIIME 2 environment (even though the environment already has iow=1.0.5 installed by default) also downgrades scikit-bio and scipy. It seems like as soon as you remind pip that iow exists, it panics and tries to adjust versions to match :)

This seems to be an artifact of QIIME 2 relying on an old iow version (as discussed on the QIIME 2 forum here).

Tentative solution

I verified that running pip install "scikit-bio==0.6.2" in the broken QIIME 2 environment seems to fix things -- at least, it allows the qiime diversity core-metrics command to run afterwards. I guess this makes sense -- this error occurs because the scikit-bio version has gotten downgraded, so fixing that version fixes the error.

However, this is not an ideal solution... ideally the environment would not get broken in the first place! But it seems like this might be due to broader QIIME 2 things (see the forum post linked above), so this might be the best we can do for now.

Steps to reproduce the issue, just for reference

• install a fresh QIIME 2 2025.7 environment

• download the DADA2 table.qza and sample-metadata.tsv from the moving pictures tutorial

• run qiime diversity core-metrics --i-table table.qza --p-sampling-depth 1103 --m-metadata-file sample-metadata.tsv --output-dir coremetrics

  • this succeeds!

• run pip install empress

• rerun the qiime diversity core-metrics command

  • this fails with the same error you got: module 'skbio.diversity.alpha' has no attribute 'sobs'