|
0
|
1 <?xml version="1.0"?>
|
|
|
2 <tool id="windowmasker_mkcounts" name="WindowMasker_mkcounts" version="1.0">
|
|
|
3 <description>Construct WindowMasker unit counts table</description>
|
|
|
4
|
|
|
5 <macros>
|
|
|
6 <import>windowmasker_macros.xml</import>
|
|
|
7 </macros>
|
|
|
8
|
|
|
9 <expand macro="requirements" />
|
|
|
10
|
|
|
11 <command detect_errors="exit_code">
|
|
|
12 <![CDATA[
|
|
|
13 @OPTIONAL_PARAM_FUNC@
|
|
|
14
|
|
|
15 windowmasker -mk_counts -infmt fasta
|
|
|
16
|
|
|
17 $optional_param("-genome_size", $adv.genome_size)
|
|
|
18 $optional_param("-t_low", $adv.t_low)
|
|
|
19 $optional_param("-t_high", $adv.t_high)
|
|
|
20 $optional_param("-unit", $adv.unit)
|
|
|
21
|
|
|
22 -sformat ${adv.sformat} -checkdup ${adv.checkdup}
|
|
|
23 -in "${fasta_input}" -out "${mkcount_output}"
|
|
|
24 ]]>
|
|
|
25 </command>
|
|
|
26 <inputs>
|
|
|
27 <param name="fasta_input" type="data" format="fasta"
|
|
|
28 label="FASTA sequence file" />
|
|
|
29
|
|
|
30 <section name="adv" title="Advanced options" expanded="false" >
|
|
|
31 <param name="checkdup" type="boolean" checked="false"
|
|
|
32 truevalue="true" falsevalue="false"
|
|
|
33 label="Check input file for potentially duplicated sequences"
|
|
|
34 help="-checkdup" />
|
|
|
35
|
|
|
36 <param name="genome_size" type="integer" label="Genome size"
|
|
|
37 min="0" optional="true"
|
|
|
38 help="Use this genome size to compute the unit length" />
|
|
|
39
|
|
|
40 <param name="sformat" type="select" label="Unit counts format"
|
|
|
41 help="The format the unit counts data should be generated" >
|
|
|
42
|
|
|
43 <option value="ascii">Plain text (ascii)</option>
|
|
|
44 <option value="oascii" selected="true">Optimized text (oascii)</option>
|
|
|
45 <!-- Omit support for non-portable binary formats (binary, obinary) -->
|
|
|
46 </param>
|
|
|
47
|
|
|
48 <param name="t_low" type="integer" label="T_low"
|
|
|
49 min="0" optional="true"
|
|
|
50 help="Save only units that appears at least this many times" />
|
|
|
51
|
|
|
52 <param name="t_high" type="integer" label="T_high"
|
|
|
53 min="0" optional="true"
|
|
|
54 help="Set the maximum count value for units" />
|
|
|
55
|
|
|
56 <param name="unit" type="integer" label="Unit length"
|
|
|
57 min="1" max="16" optional="true"
|
|
|
58 help="Unit length used to mask the genome" />
|
|
|
59 </section>
|
|
|
60 </inputs>
|
|
|
61 <outputs>
|
|
|
62 <data name="mkcount_output" format="txt" />
|
|
|
63 </outputs>
|
|
|
64 <tests>
|
|
|
65 <test>
|
|
|
66 <!-- Test mkcounts with default settings -->
|
|
|
67 <param name="fasta_input" value="chr4_part.fa" ftype="fasta" />
|
|
|
68 <output name="mkcount_output" file="chr4_part.oascii.counts" />
|
|
|
69 </test>
|
|
|
70 <test>
|
|
|
71 <!-- Test mkcounts with custom genome_size -->
|
|
|
72 <param name="fasta_input" value="chr4_part.fa" ftype="fasta" />
|
|
|
73 <param name="genome_size" value="500000" />
|
|
|
74 <output name="mkcount_output" file="chr4_part.500k_size.counts" />
|
|
|
75 </test>
|
|
|
76 <test>
|
|
|
77 <!-- Test mkcounts with ascii output format -->
|
|
|
78 <param name="fasta_input" value="chr4_part.fa" ftype="fasta" />
|
|
|
79 <param name="sformat" value="ascii" />
|
|
|
80 <output name="mkcount_output" file="chr4_part.ascii.counts" />
|
|
|
81 </test>
|
|
|
82 <test>
|
|
|
83 <!-- Test mkcounts with custom unit settings -->
|
|
|
84 <param name="fasta_input" value="chr4_part.fa" ftype="fasta" />
|
|
|
85 <param name="t_low" value="3" />
|
|
|
86 <param name="t_high" value="20" />
|
|
|
87 <param name="unit" value="9" />
|
|
|
88 <output name="mkcount_output" file="chr4_part.custom_unit.counts" />
|
|
|
89 </test>
|
|
|
90 </tests>
|
|
|
91 <help>
|
|
|
92 <![CDATA[
|
|
|
93 **What it does**
|
|
|
94
|
|
|
95 This tool runs `stage 1 <https://www.ncbi.nlm.nih.gov/IEB/ToolBox/CPP_DOC/lxr/source/src/app/winmasker/>`_
|
|
|
96 of the WindowMasker analysis to produce a unit counts file for a genome assembly.
|
|
|
97
|
|
|
98 ]]></help>
|
|
|
99
|
|
|
100 <expand macro="citations" />
|
|
|
101 </tool>
|
