Combining 16S (V4) and shotgun on one Greengenes2 tree for UniFrac — namespace mismatch (2024.09)

Library Support
1

Hello!

I am computing UniFrac across paired 16S (V4) and shotgun metagenomic data on the shared Greengenes2 phylogeny and I have ended up with three different feature ID namespaces that I can't get onto a single tree. I would appreciate guidance on which namespace the combined analysis should live in and how to convert into it.

What I have:

  • 16S table (md5). My V4 reads were paired end merged, so I denoised forward only and truncated to 150 nt to match the placed fragments. qiime greengenes2 filter-features against 2024.09.taxonomy.md5.nwk.qza then retained 100% of features. The resulting FeatureTable has md5 feature IDs.

  • Shotgun table (GG2 record IDs). Woltka/WoL2 genome IDs translated to Greengenes2 IDs (G..., GB-GCA-..., RS-GCF-...) via the WoL2-->GG2 crosswalk (wol2gg.tsv from the WoL2 release).

  • Pruned tree (asv namespace). Pruned from 2024.09.phylogeny.asv.nwk.qza, so its tips are ASV sequences with genome IDs for backbone records.

So the tree is asv-namespace, the 16S table is md5 and the shotgun side is record IDs—three namespaces and my 16S features are consequently not tips in the tree so cross platform UniFrac comes out empty.

Questions:

  1. For a combined 16S + shotgun UniFrac analysis, which namespace is the intended target and how do I get all three inputs into it?

  2. Is there a supported way to relabel an md5 filter-features table to record IDs (or ASV/sequence IDs) for 2024.09? qiime greengenes2 relabel requires a ReferenceMap artifact,

    which I can't find in the 2024.09 release and don't see an action to build. (I saw the earlier FeatureMap discussion on the 2022.10 thread and wasn't sure whether anything shipped.)

asv-table-gg2-v4.qza (45.9 KB)

asv-seqs-filtered.qza (45.0 KB)

Attached: my md5 16S feature table and the corresponding rep-seqs. Happy to provide the shotgun table or crosswalk if useful. Thanks!

4

Hi @dbadri,

If the 16S are using MD5s then the phylogeny needs to use the MD5 variant not the ASV one. The trees are the same, but the labels on the tips differ from whether they use MD5 or the ASV sequence for the tip names; the genome IDs in the tree are unaffected. I don't know what the crosswalk is that's referenced, but the genome IDs used in the phylogeny are the same ID structure in WoL2 (e.g., G#########)

Best,

Daniel

5

I downloaded the MD5 tree and it solved the problem. Thank you!