On the 29th of september SILVA published their release 128. Even though I ran into some issues last time I tried formatting the Silva release 123 by myself (SILVA v123), I (think) I figured out which errors I made last time and gave it another go.
While all runs well (I use exactly the README file provided by prof. dr. Schloss: http://blog.mothur.org/2015/12/03/SILVA-v123-reference-files/, without any deviations this time around), again the recreated seed database is smaller than before.
Mothur "#get.seqs(fasta=silva.nr_v128.align, taxonomy=silva.full_v128.tax, accnos=silva.seed_v128.accnos)"
Takes only 11213 sequences from my fasta/taxonomy file as opposed to 14914 for release 123. This worries me quite a bit. Especially since the seed is supposed to contain 70512 sequences (7* more!) (https://www.arb-silva.de/documentation/release-128/, under “New in Release 128”). This is apparently approx. 1300 sequences more than 123, which doesn’t correspond at all to my previous finds.
Did anyone run into the same issue?