See also
Rarefaction
OTU table
OTU commands
Making an OTU table (otutab command)
Generate an
OTU table by subsampling (also called rarefying). The subsampled table is
the simulated result if fewer observations (reads) had been available. The input file must be in
QIIME classic format.
The size of the subsample is specified by the -sample_pct option. E.g., if you set -sample_pct 50, this will make an OTU table assuming that half the reads had been generated.
To make a rarefaction curve for one or more alpha diversity metrics, it is easier to use the alpha_div_rare command.
The output file is specified by the -output option. The output file is also in QIIME classic format.
The method to use is specified by the -method option. Valid values are fast (the default), with_replacement and without_replacement. For all methods except fast, the number of iterations is specified by the -iters option (default 32).
Fast subsampling
Uses a fast approximation to randomized
subsampling.
With / without replacement
Subsamples using a random
number generator. Counts are averaged over several iterations and rounded to
the nearest integer. When replacement is allowed, a read may be selected
more than once. If replacement is not allowed, a read may only be selected
once.
Which method should you use?
There is some disagreement
in the literature about whether subsamping / rarefaction should be performed
with or without replacement. In my opinion, it doesn't matter because there
are much bigger issues in intepreting
counts, so I would suggest using the default fast approximation because all
three methods usually give similar results.
Example
usearch -otutab_subsample otutable.txt -subsample_pct 75 -output otutable_75.txt