In terms of the rRNA databases used for rRNA removal, I see that the default databases in the pipeline are not the same as the one recommended in SortMeRna readme. Not sure how much actual difference there is between them, but I wonder if I should use the recommended one.

We probably just haven't kept up with them. There were licensing niceties for while with silva too, though I think those are resolved. I doubt it makes a massive difference, but using the sortmerna recommendations will be fine.