|
0
|
1 <tool id="vsearch_dereplication" name="VSearch dereplication" version="@VERSION@.0">
|
|
|
2 <description></description>
|
|
|
3 <macros>
|
|
|
4 <import>vsearch_macros.xml</import>
|
|
|
5 </macros>
|
|
|
6 <expand macro="requirements" />
|
|
|
7 <expand macro="stdio" />
|
|
|
8 <expand macro="version_command" />
|
|
|
9 <command>
|
|
|
10 <![CDATA[
|
|
|
11 vsearch
|
|
|
12 @GENERAL@
|
|
|
13 --derep_fulllength $infile
|
|
|
14 #if $maxuniquesize:
|
|
|
15 --maxuniquesize $maxuniquesize
|
|
|
16 #end if
|
|
|
17 #if $minuniquesize:
|
|
|
18 --minuniquesize $minuniquesize
|
|
|
19 #end if
|
|
|
20 --output $outfile
|
|
|
21 #if $sizein:
|
|
|
22 --sizein $sizein
|
|
|
23 #end if
|
|
|
24 #if $sizeout:
|
|
|
25 --sizeout $sizeout
|
|
|
26 #end if
|
|
|
27 --strand $strand
|
|
|
28 --topn $topn
|
|
|
29 #if $uc:
|
|
|
30 --uc $uc
|
|
|
31 #end if
|
|
|
32 ]]>
|
|
|
33 </command>
|
|
|
34 <inputs>
|
|
|
35 <param name="infile" type="data" format="fasta" label="Select your FASTA file" help="(--derep_fulllength)" />
|
|
|
36 <expand macro="topn" />
|
|
|
37 <expand macro="sizein" />
|
|
|
38 <expand macro="sizeout" />
|
|
|
39 <expand macro="strand" />
|
|
|
40 <expand macro="uclust_like_output" />
|
|
|
41 <param name="minuniquesize" type="integer" value="" optional="True" label="Minimum abundance"
|
|
|
42 help="(--minuniquesize)"/>
|
|
|
43 <param name="maxuniquesize" type="integer" value="" optional="True" label="Maximum abundance"
|
|
|
44 help="(--maxuniquesize)"/>
|
|
|
45 </inputs>
|
|
|
46 <outputs>
|
|
|
47 <data name="outfile" format="fasta" label="${tool.name} on ${on_string}" />
|
|
|
48 <data name="uc" format="fasta" label="${tool.name} on ${on_string}: UCLUST like output">
|
|
|
49 <filter>uc is True</filter>
|
|
|
50 </data>
|
|
|
51 </outputs>
|
|
|
52 <tests>
|
|
|
53 <test>
|
|
|
54 <param name="infile" value="AF091148_first_rep.fsa.bz2" ftype="fasta" />
|
|
|
55 <param name="strand" value="both" />
|
|
|
56 <param name="minuniquesize" value="1" />
|
|
|
57 <param name="maxuniquesize" value="100000" />
|
|
|
58 <param name="sizeout" value="--sizeout"/>
|
|
|
59 <param name="topn" value="10000" />
|
|
|
60 <output name="outfile" file="dereplication_result1.fasta" ftype="fasta" />
|
|
|
61 </test>
|
|
|
62 </tests>
|
|
|
63 <help>
|
|
|
64 <![CDATA[
|
|
|
65 **What it does**
|
|
|
66
|
|
|
67 Merge strictly identical sequences contained in filename. Identical sequences are
|
|
|
68 defined as having the same length and the same string of nucleotides (case insensitive,
|
|
|
69 T and U are considered the same).
|
|
|
70
|
|
|
71 Dereplication options
|
|
|
72 --derep_fulllength FILENAME dereplicate sequences in the given FASTA file
|
|
|
73 --maxuniquesize INT maximum abundance for output from dereplication
|
|
|
74 --minuniquesize INT minimum abundance for output from dereplication
|
|
|
75 --output FILENAME output FASTA file
|
|
|
76 --sizein read abundance annotation from input
|
|
|
77 --sizeout write abundance annotation to output
|
|
|
78 --strand dereplicate "plus" or "both" strands (plus)
|
|
|
79 --topn INT output just the n most abundant sequences
|
|
|
80 --uc FILENAME filename for UCLUST-like output
|
|
|
81
|
|
|
82
|
|
|
83 @EXTERNAL_DOCUMENTATION@
|
|
|
84
|
|
|
85 -------
|
|
|
86
|
|
|
87 @REFERENCES@
|
|
|
88
|
|
|
89
|
|
|
90 ]]>
|
|
|
91 </help>
|
|
|
92 <expand macro="citations" />
|
|
|
93 </tool>
|
