Commands > OTU analysis
Making an OTU table (otutab command)
Remove low-abundance counts, samples and OTUs from an
OTU table. The input file must be in
QIIME classic format. The output file is specified by the -output option. It is written in QIIME classic format.
Options to set minimum abundances are min_sample_size, min_count, min_freq,
min_otu_size and min_otu_freq. See below for definitions and defaults.
Typical uses for this command include: 1. deleting samples with too few
reads, 2. creating a reduced table with the long tail of small OTUs delete
for easier processing and manual review, and 3. deleting low-abundance
counts which may be spurious due to contamination and
Deleting samples with too few reads
Setting a minimum
number of reads for a sample can be quite subjective and depends on the
objectives of your analysis. As a rule of thumb, I prefer to have at least
5,000 reads per sample. Example command line:
usearch -otutab_trim otutable.txt -min_sample_size 5000 -output trimmed.txt
Creating a reduced table for manual review
purposes, OTUs with a frequency < 1% are not very interesting or useful
because they could be spurious due to experimental error and may have
limited ecological relevance. However, it is important to keep in mind that
read abundance is a
very poor indication of species abundance, so a low-abundance OTU may in fact
have a high abundance of cells in the community). My rule of thumb is to
discard OTUs with frequency <0.5% for manual review and further analysis.
Example command line:
The threshold of 0.5% is also a
reasonable default for filtering cross-talk.
Example command line:
usearch -otutab_trim otutable.txt -min_otu_freq 0.005 -output trimmed.txt
Options which specify minimum abundances
Default 1. Samples with total count less than this are deleted.
Default 1. Minimum count. Counts less than this are set
Default 0.0. Minimum frequency, defined
as count divided by total sample size. Counts smaller than this are set to
Default 1. Minimum total size for an
OTU. OTUs smaller than this are deleted.
Default 0.0. Minimum size for an OTU as a fraction of all OTUs. OTUs smaller
than this are deleted.
If no minimums are specified, the effect is to
delete samples and OTUs which have zero.total size.