I think it would be handy to have
mafft --large option available.
Many people, including myself are taking advantage of q2 pipeline to process long 16S rRNA reads generated by Oxford Nanopore technology.
Usually mafft step aligning seqs depletes RAM and the job stops.
mafft --large option has a possibility to circumvent this problem, by saving
tmp data into files instead of using RAM.
Would that be a possibility?