We’re using the IM-TORNADO pipeline, which internally is using mothur for classify.seqs amongst other things
When we analyze some 18s amplicon data we have noticed that the classification appears off in many cases. This is using the provided Silva r119 SSU data from the mothur site (http://www.mothur.org/wiki/Silva_reference_files).
As an example, here is a subset of the OTU taxonomy results w/ Nemotoda:
...
30 Eukaryota(100);Opisthokonta(100);Holozoa(100);Metazoa(100);Animalia(100);Nematoda(100);
74 Eukaryota(100);Opisthokonta(100);Holozoa(100);Metazoa(100);Animalia(100);Nemertea(100);
297 Eukaryota(100);Opisthokonta(100);Holozoa(100);Metazoa(100);Animalia(100);Nematoda(99.9);
482 Eukaryota(100);Opisthokonta(75.6);Holozoa(75.6);Metazoa(75);Animalia(75);Nematoda(55.4);
637 Eukaryota(100);Opisthokonta(85.1);Holozoa(85.1);Metazoa(84.9);Animalia(84.9);Nematoda(70.6);
...
According to the tax_slv_ssu_nr_119.txt file obtained from the Silva downloads site (http://www.arb-silva.de/fileadmin/silva_databases/release_119/Exports/taxonomy/tax_slv_ssu_nr_119.txt) ‘Nemotoda’ is not ranked, and ‘Animalia’ is a sub-phylum (here they are ‘genus’ and ‘phylum’).
We can go back and validate the classification for our other runs (we also have 16s) and possibly fix using the above file, but I wanted to know whether anyone else has run into the same issue.