Mercurial > repos > artbio > cherry_pick_fasta

comparison cherry_pick_fasta.xml @ 3:c282a8a47dd9 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/cherry_pick_fasta commit d637de6c1090314bd34bdffc2fdf979cb55b870b"
author artbio
date Fri, 21 May 2021 09:34:14 +0000
parents 321cad0eb507
children ba6c4aeb22ea
comparison
equal deleted inserted replaced
2:321cad0eb507 3:c282a8a47dd9
1 <tool id="cherry_pick_fasta" name="Pick Fasta sequences" version="2.1.0"> 1 <tool id="cherry_pick_fasta" name="Pick Fasta sequences" version="3.0.0">
2 <description>with header satisfying a string query</description> 2 <description>with header satisfying a string query</description>
3 <requirements> 3 <requirements>
4 <requirement type="package" version="3.7.6">python</requirement> 4 <requirement type="package" version="1.70">biopython</requirement>
5 </requirements> 5 </requirements>
6 <command interpreter="python">cherry_pick_fasta.py 6 <command interpreter="python">cherry_pick_fasta.py
7 --input $input 7 --input $input
8 --searchfor '$search.searchfor' 8 --searchfor '$search.searchfor'
9 #if $search.options_selector == 'single': 9 #if $search.options_selector == 'single':
10 --query-string '$search.query' 10 #if $search.match == 'exact':
11 --query-string '$search.query' --mode exact
12 #else:
13 --query-string '$search.query' --mode includes
14 #end if
11 #else: 15 #else:
12 --query-file '$search.query' 16 #if $search.match == 'exact':
17 --query-file '$search.query' --mode exact
18 #else:
19 --query-file '$search.query' --mode includes
20 #end if
13 #end if 21 #end if
14 --output $output 22 --output $output
15 </command> 23 </command>
16 24
17 <inputs> 25 <inputs>
18 <param name="input" type="data" format="fasta" label="Source file" help="Fasta file to parse" /> 26 <param name="input" type="data" format="fasta" label="Source file" help="Fasta file to parse" />
19 27
20 <conditional name="search"> 28 <conditional name="search">
21 <param name="options_selector" type="select" display="radio" label="by single term or file of terms"> 29 <param name="options_selector" type="select" display="radio" label="for a">
22 <option value="single" selected="True">single term</option> 30 <option value="single" selected="True">single string</option>
23 <option value="textdataset">terms in a text dataset</option> 31 <option value="textdataset">list of strings</option>
24 </param> 32 </param>
25 <when value="single"> 33 <when value="single">
34 <param name="match" type="select" label="retrieve sequences whose headers...">
35 <option value="include" selected="true">partially</option>
36 <option value="exact">exactly</option>
37 </param>
38 <param name="searchfor" type="select" label=" ">
39 <option value="with" selected="true">contain this string</option>
40 <option value="without">do not contain this string</option>
41 </param>
26 <param name="query" type="text" size="30" value="" label="Search string" help="exemple: gi|40557596"> 42 <param name="query" type="text" size="30" value="" label="Search string" help="exemple: gi|40557596">
27 <sanitizer> 43 <sanitizer>
28 <valid initial="string.printable"> 44 <valid initial="string.printable">
29 <remove value="&quot;"/> 45 <remove value="&quot;"/>
30 <remove value="\"/> 46 <remove value="\"/>
33 <add source="&quot;" target="\&quot;"/> 49 <add source="&quot;" target="\&quot;"/>
34 <add source="\" target="\\"/> 50 <add source="\" target="\\"/>
35 </mapping> 51 </mapping>
36 </sanitizer> 52 </sanitizer>
37 </param> 53 </param>
38 <param name="searchfor" type="select" label="retrieve sequences whose headers contain or do not contain the search string">
39 <option value="with" selected="true">contain</option>
40 <option value="without">do not contain</option>
41 </param>
42 </when> 54 </when>
43 <when value="textdataset"> 55 <when value="textdataset">
44 <param name="query" type="data" format="txt" label="term dataset" help="a list of term to search for, one term per line" /> 56 <param name="match" type="select" label="retrieve sequences whose headers...">
45 <param name="searchfor" type="select" label="retrieve sequences whose headers contain or do not contain the search list"> 57 <option value="includes" selected="true">partially</option>
46 <option value="with" selected="true">contain</option> 58 <option value="exact">exactly</option>
47 <option value="without">do not contain</option>
48 </param> 59 </param>
60 <param name="searchfor" type="select" label=" ">
61 <option value="with" selected="true">contain one of these list strings</option>
62 <option value="without">do not contain one of these list strings</option>
63 </param>
64 <param name="query" type="data" format="txt" label="list of strings dataset" help="a list of strings to search for, one string per line" />
49 </when> 65 </when>
50 </conditional> 66 </conditional>
51 </inputs> 67 </inputs>
52 <outputs> 68 <outputs>
53 <data name="output" format="fasta" label="Fasta sequences ${search.searchfor.value} ${search.options_selector} term(s) in header" /> 69 <data name="output" format="fasta" label="Fasta sequences ${search.searchfor.value} ${search.options_selector} term(s) in header" />
54 </outputs> 70 </outputs>
55 <tests> 71 <tests>
72 <!-- exact matches -->
56 <test> 73 <test>
57 <param ftype="fasta" name="input" value="input.fa" /> 74 <param ftype="fasta" name="input" value="input.fa" />
58 <!-- <param name="options_selector" value="textdataset" /> --> 75 <param name="query" value="gi|81971654|sp|Q9IJX4.1|POLN_CRPVC_RecName:_Full_Replicase_polyprotein;_Contains:_RecName:_Full_Pro--Locus_65_Transcript_1/2_Confidence_0.667_Length_1344_hit1_IdMatch=43.46,AligLength=451,E-val=2e-122" />
76 <param name="searchfor" value="without" />
77 <param name="match" value="exact" />
78 <output name="output" ftype="fasta" file="output_exactly_not.fa" />
79 </test>
80 <test>
81 <param ftype="fasta" name="input" value="input.fa" />
82 <param name="query" value="gi|81971654|sp|Q9IJX4.1|POLN_CRPVC_RecName:_Full_Replicase_polyprotein;_Contains:_RecName:_Full_Pro--Locus_65_Transcript_1/2_Confidence_0.667_Length_1344_hit1_IdMatch=43.46,AligLength=451,E-val=2e-122" />
83 <param name="searchfor" value="with" />
84 <param name="match" value="exact" />
85 <output name="output" ftype="fasta" file="output_exact.fa" />
86 </test>
87
88
89 <test>
90 <param ftype="fasta" name="input" value="input.fa" />
91 <param name="options_selector" value="textdataset" />
92 <param name="query" ftype="txt" value="alt_termlist.txt" />
93 <param name="searchfor" value="without" />
94 <param name="match" value="exact" />
95 <output name="output" ftype="fasta" file="output_alt_termlist_without.fa" />
96 </test>
97 <test>
98 <param ftype="fasta" name="input" value="input.fa" />
99 <param name="options_selector" value="textdataset" />
100 <param name="query" ftype="txt" value="alt_termlist.txt" />
101 <param name="searchfor" value="with" />
102 <param name="match" value="exact" />
103 <output name="output" ftype="fasta" file="output_alt_termlist.fa" />
104 </test>
105
106
107 <!-- partial matches -->
108 <test>
109 <param ftype="fasta" name="input" value="input.fa" />
59 <param name="query" value="gi|81971654" /> 110 <param name="query" value="gi|81971654" />
60 <param name="searchfor" value="with" /> 111 <param name="searchfor" value="with" />
61 <output name="output" ftype="fasta" file="output.fa" /> 112 <output name="output" ftype="fasta" file="output.fa" />
62 </test> 113 </test>
63 <test> 114 <test>
64 <param ftype="fasta" name="input" value="input.fa" /> 115 <param ftype="fasta" name="input" value="input.fa" />
65 <!-- <param name="options_selector" value="textdataset" /> -->
66 <param name="query" value="RNA" /> 116 <param name="query" value="RNA" />
67 <param name="searchfor" value="without" /> 117 <param name="searchfor" value="without" />
68 <output name="output" ftype="fasta" file="output_without.fa" /> 118 <output name="output" ftype="fasta" file="output_without.fa" />
69 </test> 119 </test>
70 <test> 120 <test>
80 <param name="query" ftype="txt" value="termlist.txt" /> 130 <param name="query" ftype="txt" value="termlist.txt" />
81 <param name="searchfor" value="without" /> 131 <param name="searchfor" value="without" />
82 <output name="output" ftype="fasta" file="output_termlist_without.fa" /> 132 <output name="output" ftype="fasta" file="output_termlist_without.fa" />
83 </test> 133 </test>
84 </tests> 134 </tests>
85
86 <help> 135 <help>
87 **What it does** 136 **What it does**
88 137
89 This tool retrieves nucleotide/peptide sequences from a fasta file whose headers match 138 This tool retrieves nucleotide/peptide sequences from a fasta file whose headers match
90 or do not match a given string. 139 or do not match a given string, or a list of strings.
91
92 It is Copyright © 2019 `CNRS and Sorbonne-Université`_ and is released under the `MIT license`_.
93
94 .. _CNRS and Sorbonne-Université: http://www.sorbonne-universite.fr/en
95 .. _MIT license: http://opensource.org/licenses/MIT
96 140
97 </help> 141 </help>
98 </tool> 142 </tool>