feature-table summarize odd


in qiime2-amplicon-2024.2-py38-linux-conda the output of feature-table summarize on a denoised feature table that has been generated by merging individual denoised runs and then same samples from different runs grouped with feature-table group, is a bit odd on the interactive sample detail. First off, it doesn't show the individual sample graph, and changing the sampling depth gives odd number as in:

Retained 25,252,804 (462.19%) features in 212 (100.00%) samples at the specifed sampling depth.

Also, the samples are labeled oddly with numbers 0 to (max sample number) instead of the actual IDs, regardless if a metadata file is provided or not. Changing the slider changes the number of retained features but the number of samples always stays at 100% which is not possible at a sampling depth (e.g. ~10000) that obviously has more features than the samples with less total features (e.g. 4462), as in the screenshot:

Compared to a summarized table .qzv generated in an older qiime environment (of a different dataset), where the number of retained features never was >100% and changing the sampling depth reduced the number of reported samples retaining features as expected:

Attached the table qzv in question (after lots of postprocessing).
dada2-table-grouped_decont-nocont-min2-noMitChl2-noUna.qzv (736.9 KB)

Any table summary created by feature-table summarize in this environment seems to produce the same bugged qzv.
When I create other summaries from the same feature table, e.g. taxa barplot, the sample id labels are all normally reported in the qzv and the graphs.


1 Like

Hi @lxsteiner,
This was a known bug in feature table summarize that was patched later on. I recommend updating your qiime environment to the latest release. I think this will resolve the issue you are seeing here!

1 Like

Thanks @jphagen good to know.

1 Like