wasade
commented
1 year ago Hi @chrysn,
Thank you for opening this issue, and I'm sorry to learn about the inconsistency. On the surface, it sounds like there may be a different interpretation of the format. This repository is the official BIOM implementation, and the format specification the HDF5 container can be found here.
If you do decide to open an issue with rbiom, could you tag me and cross reference this issue? Cross language interoperability is extremely important, and while I'm unaffiliated with rbiom, I'd be eager to work with them to normalize processing.
Exchanging data in the biom format between Python (using the biom-format package) and R (using rbiom) is using different and incompatible formats:
On the Python side, taxonomy is typically stored in metadata on the observation axis (eg. in test_cli/test_summarize_table.py:
"metadata": {"taxonomy": ["k__Bacteria"]}).That data appears to be unavailable when loading the biom file through rbiom.
From R, taxonomy data is feature of its own, and data stored in
biom$taxonomyappears not to be exposed when loading the biom file in Python.(All my tests I've done with the h5 storage format.)
I'm not sure whether this is a matter of implementation shortcomings or lack of clarity in the specification; if it is the former (and the rbiom package is the one that got it wrong), I'd appreciate any pointer inside the specification I could use when opening a replacement issue on the rbiom side.