Closed-reference OTU assignment (Rideout
et al. 2014) assigns query sequences to OTUs by searching a
pre-defined database of full-length sequences which have been clustered at
97% identity. In QIIME, it is implemented by the
pick_closed_reference_otus.py script using a
default database which was obtained by
clustering Greengenes. In QIIME v1.9, the database search is performed by
uclust, an old software package that was the predecessor of usearch.
In usearch, a similar algorithm is implemented in the
closed_ref command. The
USEARCH algorithm is used to search the
database. Different parameters are used compared to the
usearch_global command to improve
sensitivity and report cases where two or more database sequences are tied
for the highest identity. Ties are broken systematically by reporting the
first of hit in database file order.