A few notes on 16S analysis
05 Nov 2013- You can use MEGAN to import TAB files exported by USEARCH, but only if you use USEARCH in local mode, as otherwise you don’t get e-values which MEGAN doesn’t like, e.g.:
-
USEARCH 32-bit mode (the free version) has a 4GB memory limit, which later versions of SILVA can exceed. SILVA 111 NR fits within 2B.
-
MEGAN has a SILVA to GI lookup file, but this is only for version 108 of SILVA. Not clear how to make your own one from ARB files without writing some kind of bodge look-up script.
-
In order to crunch tab files from the command-line you need something like:
-
sleepcat.py is just a script to make this process more reliable
-
You need -maxrejects 0 -maxaccepts 0 to export multiple hits so that MEGAN will work with LCA mode (USEARCH just returns the top hit by default)