I finally got in touch with someone at ENA, and it seems a bigger problem than I thought.
So, by mistake, I submitted the MAG assemblies to the soil biosamples they came from, not the MAG biosamples. The issue is that the GCA accessions are generated based on sample and project accnos combination. You cannot change the assembly metadata, e.g. taxonomy, that has to change in the sample:
Once an assembly is submitted to the ENA, it is not possible to change the
sample accession (ERSxxx) linked to that assembly, this is because the GCA is
generated based on the sample and project accession combination. It is only
possible to change the taxonomy metadata (scientific_name and tax_id) in the
sample xml for the original ERS and that update can then be reflected in the
assembly records. Analysis xmls are largely uneditable.
They told me that they had to talk with the backend devs to see if this was doable as it’s something they definitely don’t do, but they are also worrying about potential issues with the data exchange with the other INSDC databases…
My case is not the only case (see Benthobacter in Spongiisociales appears with a placeholder in GTDB226). Wondering if the GTDB couldn’t pull the accessions from the SeqCode ratified species records considering possible cases like these and compare the accnos to the NCBI’s assemblies, esp. since they are also genomes that are thoroughly checked.