What it does

vsearch implements a single-pass, greedy star-clustering algorithm, similar to the algorithms implemented in usearch, DNAclust and sumaclust for example.

Clustering options (most searching options also apply)

--centroids FILENAME
	output centroid sequences to FASTA file
--cluster_fast FILENAME
	cluster sequences after sorting by length
--cluster_size FILENAME
	cluster sequences after sorting by abundance
--cluster_smallmem FILENAME
	cluster already sorted sequences (see -usersort)
--clusters STRING
	output each cluster to a separate FASTA file
--consout FILENAME
	output cluster consensus sequences to FASTA file
--cons_truncate
	do not ignore terminal gaps in MSA for consensus
--id REAL	reject if identity lower
--iddef INT	id definition, 0-4=CD-HIT,all,int,MBL,BLAST (2)
--msaout FILENAME
	output multiple seq. alignments to FASTA file
--qmask	seqs with dust, soft or no method (dust)
--sizein	propagate abundance annotation from input
--sizeout	write cluster abundances to centroid file
--strand	cluster using plus or both strands (plus)
--uc FILENAME	filename for UCLUST-like output
--usersort	indicate sequences not presorted by length

For details about this tool, please refer to the GitHub repository or the vsearch manual.