Mercurial > repos > artbio > cherry_pick_fasta
diff cherry_pick_fasta.xml @ 3:c282a8a47dd9 draft
"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 |
line wrap: on
line diff
--- a/cherry_pick_fasta.xml Tue Mar 16 23:25:57 2021 +0000 +++ b/cherry_pick_fasta.xml Fri May 21 09:34:14 2021 +0000 @@ -1,15 +1,23 @@ -<tool id="cherry_pick_fasta" name="Pick Fasta sequences" version="2.1.0"> +<tool id="cherry_pick_fasta" name="Pick Fasta sequences" version="3.0.0"> <description>with header satisfying a string query</description> <requirements> - <requirement type="package" version="3.7.6">python</requirement> + <requirement type="package" version="1.70">biopython</requirement> </requirements> <command interpreter="python">cherry_pick_fasta.py --input $input --searchfor '$search.searchfor' #if $search.options_selector == 'single': - --query-string '$search.query' + #if $search.match == 'exact': + --query-string '$search.query' --mode exact + #else: + --query-string '$search.query' --mode includes + #end if #else: - --query-file '$search.query' + #if $search.match == 'exact': + --query-file '$search.query' --mode exact + #else: + --query-file '$search.query' --mode includes + #end if #end if --output $output </command> @@ -18,11 +26,19 @@ <param name="input" type="data" format="fasta" label="Source file" help="Fasta file to parse" /> <conditional name="search"> - <param name="options_selector" type="select" display="radio" label="by single term or file of terms"> - <option value="single" selected="True">single term</option> - <option value="textdataset">terms in a text dataset</option> + <param name="options_selector" type="select" display="radio" label="for a"> + <option value="single" selected="True">single string</option> + <option value="textdataset">list of strings</option> </param> <when value="single"> + <param name="match" type="select" label="retrieve sequences whose headers..."> + <option value="include" selected="true">partially</option> + <option value="exact">exactly</option> + </param> + <param name="searchfor" type="select" label=" "> + <option value="with" selected="true">contain this string</option> + <option value="without">do not contain this string</option> + </param> <param name="query" type="text" size="30" value="" label="Search string" help="exemple: gi|40557596"> <sanitizer> <valid initial="string.printable"> @@ -35,17 +51,17 @@ </mapping> </sanitizer> </param> - <param name="searchfor" type="select" label="retrieve sequences whose headers contain or do not contain the search string"> - <option value="with" selected="true">contain</option> - <option value="without">do not contain</option> - </param> </when> <when value="textdataset"> - <param name="query" type="data" format="txt" label="term dataset" help="a list of term to search for, one term per line" /> - <param name="searchfor" type="select" label="retrieve sequences whose headers contain or do not contain the search list"> - <option value="with" selected="true">contain</option> - <option value="without">do not contain</option> + <param name="match" type="select" label="retrieve sequences whose headers..."> + <option value="includes" selected="true">partially</option> + <option value="exact">exactly</option> </param> + <param name="searchfor" type="select" label=" "> + <option value="with" selected="true">contain one of these list strings</option> + <option value="without">do not contain one of these list strings</option> + </param> + <param name="query" type="data" format="txt" label="list of strings dataset" help="a list of strings to search for, one string per line" /> </when> </conditional> </inputs> @@ -53,16 +69,50 @@ <data name="output" format="fasta" label="Fasta sequences ${search.searchfor.value} ${search.options_selector} term(s) in header" /> </outputs> <tests> + <!-- exact matches --> + <test> + <param ftype="fasta" name="input" value="input.fa" /> + <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" /> + <param name="searchfor" value="without" /> + <param name="match" value="exact" /> + <output name="output" ftype="fasta" file="output_exactly_not.fa" /> + </test> + <test> + <param ftype="fasta" name="input" value="input.fa" /> + <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" /> + <param name="searchfor" value="with" /> + <param name="match" value="exact" /> + <output name="output" ftype="fasta" file="output_exact.fa" /> + </test> + + <test> <param ftype="fasta" name="input" value="input.fa" /> - <!-- <param name="options_selector" value="textdataset" /> --> + <param name="options_selector" value="textdataset" /> + <param name="query" ftype="txt" value="alt_termlist.txt" /> + <param name="searchfor" value="without" /> + <param name="match" value="exact" /> + <output name="output" ftype="fasta" file="output_alt_termlist_without.fa" /> + </test> + <test> + <param ftype="fasta" name="input" value="input.fa" /> + <param name="options_selector" value="textdataset" /> + <param name="query" ftype="txt" value="alt_termlist.txt" /> + <param name="searchfor" value="with" /> + <param name="match" value="exact" /> + <output name="output" ftype="fasta" file="output_alt_termlist.fa" /> + </test> + + + <!-- partial matches --> + <test> + <param ftype="fasta" name="input" value="input.fa" /> <param name="query" value="gi|81971654" /> <param name="searchfor" value="with" /> <output name="output" ftype="fasta" file="output.fa" /> </test> <test> <param ftype="fasta" name="input" value="input.fa" /> - <!-- <param name="options_selector" value="textdataset" /> --> <param name="query" value="RNA" /> <param name="searchfor" value="without" /> <output name="output" ftype="fasta" file="output_without.fa" /> @@ -82,17 +132,11 @@ <output name="output" ftype="fasta" file="output_termlist_without.fa" /> </test> </tests> - <help> **What it does** This tool retrieves nucleotide/peptide sequences from a fasta file whose headers match -or do not match a given string. - -It is Copyright © 2019 `CNRS and Sorbonne-Université`_ and is released under the `MIT license`_. - -.. _CNRS and Sorbonne-Université: http://www.sorbonne-universite.fr/en -.. _MIT license: http://opensource.org/licenses/MIT +or do not match a given string, or a list of strings. </help> </tool>