|
0
|
1 <tool id="clc_assembler" name="CLC assembler" version="0.0.2">
|
|
|
2 <description>Assembles reads giving a FASTA file</description>
|
|
|
3 <requirements>
|
|
|
4 <requirement type="binary">clc_assembler</requirement>
|
|
|
5 </requirements>
|
|
|
6 <version_command>\${CLC_ASSEMBLY_CELL:-/mnt/apps/clcBio/clc-assembly-cell-4.1.0-linux_64/}clc_assembler | grep -i version</version_command>
|
|
|
7 <command>\${CLC_ASSEMBLY_CELL:-/mnt/apps/clcBio/clc-assembly-cell-4.1.0-linux_64/}clc_assembler
|
|
|
8 #for $rg in $read_group
|
|
|
9 ##--------------------------------------
|
|
|
10 #if str($rg.segments.type) == "paired"
|
|
|
11 -p $rg.segments.placement $rg.segments.dist_mode $rg.segments.min_size $rg.segments.max_size -q -i "$rg.segments.filename1" "$rg.segments.filename2"
|
|
|
12 #end if
|
|
|
13 ##--------------------------------------
|
|
|
14 #if str($rg.segments.type) == "interleaved"
|
|
|
15 -p $rg.segments.placement $rg.segments.dist_mode $rg.segments.min_size $rg.segments.max_size -q "$rg.segments.filename"
|
|
|
16 #end if
|
|
|
17 ##--------------------------------------
|
|
|
18 #if str($rg.segments.type) == "none"
|
|
|
19 -p no -q
|
|
|
20 #for $f in $rg.segments.filenames
|
|
|
21 "$f"
|
|
|
22 #end for
|
|
|
23 #end if
|
|
|
24 ##--------------------------------------
|
|
|
25 #end for
|
|
|
26 -m $min_contig_len
|
|
|
27 -o "$out_fasta"
|
|
|
28 --cpus \${GALAXY_SLOTS:-4}
|
|
|
29 -v | grep -v "^Progress: "</command>
|
|
|
30 <stdio>
|
|
|
31 <!-- Assume anything other than zero is an error -->
|
|
|
32 <exit_code range="1:" />
|
|
|
33 <exit_code range=":-1" />
|
|
|
34 </stdio>
|
|
|
35 <inputs>
|
|
|
36 <repeat name="read_group" title="Read Group" min="1">
|
|
|
37 <conditional name="segments">
|
|
|
38 <param name="type" type="select" label="Are these paired reads?">
|
|
|
39 <option value="paired">Paired reads (as two files)</option>
|
|
|
40 <option value="interleaved">Paired reads (as one interleaved file)</option>
|
|
|
41 <option value="none">Unpaired reads (single or orphan reads)</option>
|
|
|
42 </param>
|
|
|
43 <when value="paired">
|
|
|
44 <param name="placement" type="select" label="Pairing type (segment placing)">
|
|
|
45 <option value="fb">---> <--- (e.g. Sanger capillary or Solexa/Illumina paired-end library)</option>
|
|
|
46 <option value="bf"><--- ---> (e.g. Solexa/Illumina mate-pair library)</option>
|
|
|
47 <option value="ff">---> ---></option>
|
|
|
48 <option value="bb"><--- <---</option>
|
|
|
49 </param>
|
|
|
50 <param name="dist_mode" type="select" label="How is the fragment distance measured?">
|
|
|
51 <option value="ss">Start to start (e.g. Sanger capillary or Solexa/Illumina libraries)</option>
|
|
|
52 <option value="se">Start to end</option>
|
|
|
53 <option value="es">End to start</option>
|
|
|
54 <option value="ee">End to end</option>
|
|
|
55 </param>
|
|
|
56 <!-- TODO - min/max validation done via the <code> tag? -->
|
|
|
57 <param name="min_size" type="integer" optional="false" min="0" value=""
|
|
|
58 label="Minimum size of 'good' DNA templates in the library preparation" />
|
|
|
59 <param name="max_size" type="integer" optional="false" min="0" value=""
|
|
|
60 label="Maximum size of 'good' DNA templates in the library preparation" />
|
|
|
61 <param name="filename1" type="data" format="fastq,fasta" required="true" label="Read file one"/>
|
|
|
62 <param name="filename2" type="data" format="fastq,fasta" required="true" label="Read file two"/>
|
|
|
63 </when>
|
|
|
64 <when value="interleaved">
|
|
|
65 <param name="placement" type="select" label="Pairing type (segment placing)">
|
|
|
66 <option value="fb">---> <--- (e.g. Sanger capillary or Solexa/Illumina paired-end library)</option>
|
|
|
67 <option value="bf"><--- ---> (e.g. Solexa/Illumina mate-pair library)</option>
|
|
|
68 <option value="ff">---> ---></option>
|
|
|
69 <option value="bb"><-- <--</option>
|
|
|
70 </param>
|
|
|
71 <param name="dist_mode" type="select" label="How is the fragment distance measured?">
|
|
|
72 <option value="ss">Start to start (e.g. Sanger capillary or Solexa/Illumina libraries)</option>
|
|
|
73 <option value="se">Start to end</option>
|
|
|
74 <option value="es">End to start</option>
|
|
|
75 <option value="ee">End to end</option>
|
|
|
76 </param>
|
|
|
77 <!-- TODO - min/max validation done via the <code> tag? -->
|
|
|
78 <param name="min_size" type="integer" optional="false" min="0" value=""
|
|
|
79 label="Minimum size of 'good' DNA templates in the library preparation" />
|
|
|
80 <param name="max_size" type="integer" optional="false" min="0" value=""
|
|
|
81 label="Maximum size of 'good' DNA templates in the library preparation" />
|
|
|
82 <param name="filename" type="data" format="fastq,fasta" required="true" label="Interleaved read file"/>
|
|
|
83 </when>
|
|
|
84 <when value="none">
|
|
|
85 <param name="filenames" type="data" format="fastq,fasta" multiple="true" required="true" label="Read file(s)"
|
|
|
86 help="Multiple files allowed, for example several files of orphan reads." />
|
|
|
87 </when>
|
|
|
88 </conditional>
|
|
|
89 </repeat>
|
|
|
90 <param name="min_contig_len" type="integer" optional="false" min="1" value="200" label="Minimum contig length"/>
|
|
|
91 <!-- Word size? -->
|
|
|
92 <!-- Bubble size? -->
|
|
|
93 <!-- Scaffolding options? -->
|
|
|
94 <!-- AGP / GFF output? -->
|
|
|
95 </inputs>
|
|
|
96 <!-- min/max validation? <code file="clc_validator.py" /> -->
|
|
|
97 <outputs>
|
|
|
98 <data name="out_fasta" format="fasta" label="CLCbio assember contigs (FASTA)" />
|
|
|
99 </outputs>
|
|
|
100 <tests>
|
|
|
101 <test>
|
|
|
102 <param name="read_group_0|segments|type" value="interleaved" />
|
|
|
103 <param name="read_group_0|segments|placement" value="fb" />
|
|
|
104 <param name="read_group_0|segments|dist_mode" value="ss" />
|
|
|
105 <param name="read_group_0|segments|min_size" value="1" />
|
|
|
106 <param name="read_group_0|segments|max_size" value="1000" />
|
|
|
107 <param name="read_group_0|segments|dist_mode" value="ss" />
|
|
|
108 <param name="read_group_0|segments|filename" value="SRR639755_mito_pairs.fastq.gz" ftype="fastqsanger" />
|
|
|
109 <param name="min_contig_len" value="200" />
|
|
|
110 <output name="out_fasta" file="SRR639755_mito_pairs.clc4_de_novo.fasta" ftype="fasta" />
|
|
|
111 </test>
|
|
|
112 </tests>
|
|
|
113 <help>
|
|
|
114
|
|
|
115 **What it does**
|
|
|
116
|
|
|
117 Runs the ``clc_assembler`` tool giving a FASTA output file. You would then
|
|
|
118 typically map the same set of reads onto this assembly using ``cls_mapper``
|
|
|
119 to any perform downstream analysis using the mapped reads.
|
|
|
120
|
|
|
121
|
|
|
122 **Citation**
|
|
|
123
|
|
|
124 If you use this Galaxy tool in work leading to a scientific publication please
|
|
|
125 cite this wrapper as:
|
|
|
126
|
|
|
127 Peter J.A. Cock (2013), Galaxy wrapper for the CLC Assembly Cell suite from CLCbio
|
|
|
128 http://toolshed.g2.bx.psu.edu/view/peterjc/clc_assembly_cell
|
|
|
129
|
|
|
130 This wrapper is available to install into other Galaxy Instances via the Galaxy
|
|
|
131 Tool Shed at http://toolshed.g2.bx.psu.edu/view/peterjc/clc_assembly_cell
|
|
|
132 </help>
|
|
|
133 </tool>
|
