# chimera.vsearch

The chimera.vsearch command reads a fasta file and reference file or a fasta and name or count file and outputs potentially chimeric sequences. The vsearch program is donated to the public domain, https://github.com/torognes/vsearch.

## Default Settings

The fasta parameter is required.

mothur > chimera.vsearch(fasta=stool.trim.unique.good.align, reference=silva.gold.align)


or

mothur > chimera.vsearch(fasta=fasta=stool.trim.unique.good.align, name=fasta=stool.trim.unique.good.names)


or

mothur > chimera.vsearch(fasta=fasta=stool.trim.unique.good.align, name=fasta=stool.trim.unique.good.names, group=stool.trim.unique.good.groups)


The chimeras file format is explained here.

## Options

### vsearch

The vsearch parameter allows you to specify the name and location of your vsearch executable. By default mothur will look in your path and mothur’s executable location. You can set the vsearch location as follows: vsearch=/usr/bin/vsearch.

mothur > chimera.vsearch(vsearch=/usr/bin/vsearch.2.11.1,  fasta=stool.trim.unique.good.align, name=stool.trim.good.names)


### name

You can provide a name file to check for chimeras using more abundant sequences as the reference.

mothur > chimera.vsearch(fasta=stool.trim.unique.good.align, name=stool.trim.good.names)


### group

If you are using reference=self and provide a groupfile, mothur will use the more abundant sequences from the same sample to check the query sequence.

mothur > chimera.vsearch(fasta=stool.trim.unique.good.align, name=stool.trim.good.names, group=stool.good.groups)


### count

The count file is similar to the name file in that it is used to represent the number of duplicate sequences for a given representative sequence. The count file can also contain group information.

mothur > make.table(name=stool.trim.good.names, group=stool.good.groups)
mothur > chimera.vsearch(fasta=stool.trim.unique.good.align, count=stool.trim.good.count_table)


### processors

The processors parameter allows you to specify how many processors you would like to use. The default is 1.

### dereplicate

The dereplicate parameter can be used when checking for chimeras by group. If the dereplicate parameter is false, then if one group finds the sequence to be chimeric, then all groups find it to be chimeric, default=f. If you set dereplicate=t, and then run remove.seqs with dups=f you can remove only the redundant chimeric sequences.

Let’s look at an example:

>seq1
attgacat....
>seq4
ttgacaga....

seq1 seq1,seq2,seq3
seq4 seq4,seq5,seq6

seq1 group1
seq2 group2
seq3 group3
seq4 group1
seq5 group2
seq6 group3


If dereplicate=f and dups=t, (default settings in mothur), and seq2 is found to be chimeric by group2. The results would be:

>seq4
ttgacaga....

seq4 seq4,seq5,seq6

seq4 group1
seq5 group2
seq6 group3


If dereplicate=t and dups=t, and seq2 is found to be chimeric by group2. The results would be:

>seq4
ttgacaga....

seq4 seq4,seq5,seq6

seq4 group1
seq5 group2
seq6 group3


If dereplicate=t and dups=f, and seq2 is found to be chimeric by group2. The results would be:

>seq1
attgacat....
>seq4
ttgacaga....

seq1 seq1,seq3
seq4 seq4,seq5,seq6

seq1 group1
seq3 group3
seq4 group1
seq5 group2
seq6 group3


### abskew

The abskew parameter can only be used with template=self. Minimum abundance skew. Default 1.9. Abundance skew is: min [ abund(parent1), abund(parent2) ] / abund(query).

### chimealns

The chimealns parameter allows you to indicate you would like a file containing multiple alignments of query sequences to parents in human readable format. Alignments show columns with differences that support or contradict a chimeric model.

### minh

The minh parameter - mininum score to report chimera. Default 0.3. Values from 0.1 to 5 might be reasonable. Lower values increase sensitivity but may report more false positives. If you decrease xn you may need to increase minh, and vice versa.

### mindiv

The mindiv parameter - minimum divergence ratio, default 0.5. Div ratio is 100%% - %%identity between query sequence and the closest candidate for being a parent. If you don’t care about very close chimeras, then you could increase mindiv to, say, 1.0 or 2.0, and also decrease minh, say to 0.1, to increase sensitivity. How well this works will depend on your data. Best is to tune parameters on a good benchmark.

### mindiffs

The mindiffs parameter - minimum number of differences in segment Default = (3).

### xn

The xn parameter - weight of a no vote. Default 8.0. Decreasing this weight to around 3 or 4 may give better performance on denoised data.

### dn

The dn parameter - pseudo-count prior on number of no votes. Default

1.4. Probably no good reason to change this unless you can retune to a good benchmark for your data. Reasonable values are probably in the range from 0.2 to 2.

## Revisions

• 1.38.0 First Introduced
• 1.39.0 Adds chimera.vsearch for Windows users
• 1.39.2 Bug Fix: removing last character of sequence names when processing with a reference. Not an issue with denovo method.
• 1.40.0 Fixes screen output. #309
• 1.42.0 Updates vsearch version to 2.11.1 #585
• 1.42.0 Adds vsearch parameter to chimera.vsearch so that you can specify location of vsearch executable. #586
• 1.44.0 Adds parallelization to chimera.vsearch when using denovo method. #700