I would like to create a new training set including sequences from SILVA_138.1_SSURef_NR99_tax_silva.fasta using LearnTaxa however I think I am running into some issues with the silva taxonomy naming structure.

For LearnTaxa can the taxonomy contain symbols (ie "-" or "[")? Are blanks alright, ie " incertae sedis" vs "_incertae_sedis"?

I am curious how the taxonomy from silva was corrected for the previously trained databases (ie SILVA_SSU_r132_March2018.RData mentioned in https://benjjneb.github.io/dada2/tutorial.html - i cannot seem to find more recent trained db on the Decipher webpage right now). I have tried using both the taxonomy supplied in SILVA_138.1_SSURef_NR99_tax_silva.fasta and from taxmap_slv_ssu_ref_nr_138.1.txt , both of which have some issues. Any insight or suggestions would be appreciated!