The taxonomic assignments you are showing are in the GreenGenes format (they use the k__ prefixes for different ranks), so I assume you are actually using GreenGenes2 as your reference database.

GreenGenes2 has many entries with "blank" taxonomy in the middle ranks, so what you are seeing is expected when using that reference. I do not know the details of how they decide to do this. You could perhaps ask the GG2 developers.

I am sure I used SILVA 138, here is phyloseq object and tax table for reference (all zipped). My study is with V3-V4 primers.

phyloseq_tax.zip

The taxonomic strings that have been assigned are in GG format (maybe GreenGenes 1), not Silva format. See below:

seqs <- getSequences(system.file("extdata", "example_seqs.fa", package="dada2"))

# Assignment with GG2
assignTaxonomy(seqs, "~/tax/gg2_2024_09_toGenus_trainset.fa.gz") |> unname()

[1,] "d__Bacteria" "p__Bacillota_I" "c__Bacilli_A" "o__Lactobacillales" "f__Lactobacillaceae" "g__Secundilactobacillus"
[2,] "d__Bacteria" "p__Cyanobacteriota" "c__Chloroplast" NA NA NA
[3,] "d__Bacteria" "p__Bacillota_I" "c__Bacilli_A" "o__Bacillales_D" "f__Amphibacillaceae" NA
[4,] "d__Bacteria" "p__Patescibacteria" "c__Saccharimonadia" "o__Saccharimonadales" NA NA
[5,] "d__Bacteria" "p__Bacillota_A_368345" "c__Clostridia_258483" "o__Clostridiales" "f__Clostridiaceae_222000" "g__Clostridium_Z"
[6,] "d__Archaea" "p__Thermoproteota" NA NA NA NA

# Assignment with Silva
assignTaxonomy(seqs, "~/tax/silva_nr99_v138.2_toGenus_trainset.fa.gz") |> unname()

[1,] "Bacteria" "Bacillota" "Bacilli" "Lactobacillales" "Lactobacillaceae" "Secundilactobacillus"
[2,] "Bacteria" "Cyanobacteriota" "Cyanobacteriia" "Chloroplast" NA NA
[3,] "Bacteria" "Bacillota" "Bacilli" "Bacillales" "Bacillaceae" "Lentibacillus"
[4,] "Bacteria" "Patescibacteria" "Saccharimonadia" "Saccharimonadales" NA NA
[5,] "Bacteria" "Bacillota" "Clostridia" "Clostridiales" "Clostridiaceae" "Clostridium"
[6,] NA NA NA NA NA NA

Thanks for the info but this is not my workflow of using greengenes, I used SILVA as below:

# STEP 9. Assign Taxonomy ---- 
#for silva from kingdom to spp.
taxa <- assignTaxonomy(seqtab.nochim2, "silva_nr99_v138_train_set.fa", multithread = TRUE)
taxa <- addSpecies(taxa, "silva_species_assignment_v138.fa")

I don't know what to tell you then. The taxonomic strings you are getting do not come from our release of the Silva reference files. You can download our released files with the names you are using here: https://zenodo.org/records/3986799

If you grep e.g. Phreatobacter you won't find taxonomic id lines that contain g__Phreatobacter, or any taxonomic id lines that use the x__ prefixes that are associated with greengenes.

Maybe you have a custom formatting of Silva?

Thanks for your response. This is worrying as I already published a paper with the is workflow using this SILVA 138. So the SILVA I attached here is not your SILVA?
https://drive.google.com/file/d/1NXVGxZerwdbyoJdhRqPRA0WakUAGqb03/view?usp=sharing

The file you linked is the same as the file we released, available at the link I posted previously. You can check this quickly by calculating their checksums, e.g. md5 silva_nr99_v138_train_set.fa.gz on both versions.

Your issue is that wasn't the reference file that produced the taxonomic assignments you are looking at.