Which database should I use to run `humann_regroup_table` and `humann_rename_table` command?

Thanks a lot @lauren.j.mciver for your kind response. I have done the grouping step. I found relative abundance of 0.65-0.7 for UNMAPPED + UNGROUPED when used uniref90_go. Relative abundance for UNMAPPED ranges between 0.45-0.55 and for UNGROUPED it is around 0.2.
Do you think such level of relative abundance for the two groups (UNMAPPED and UNGROUPED) are fine? If not, what should be the normal range?
I have also noticed that the relative abundance shows only upto 3 levels of precision. As a result, I am getting too many zero abundances for many terms. I have also asked in this thread and yet to get a response. Can you please help regarding this?

Many thanks ,
DC7