Mercurial > repos > cpt > cpt_search_file

comparison searchFile.xml @ 1:6e3a843b6304 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c
author cpt
date Mon, 05 Jun 2023 02:53:18 +0000
parents
children 21ae0e340d80
comparison
equal deleted inserted replaced
0:bd2ff2c7e806 1:6e3a843b6304
1 <tool id="edu.tamu.cpt.proximity.searchFile" name="Search File" version="1.0">
2 <description>Queries a gff3, genbank, fasta, or blastxml file for a user defined set of terms</description>
3 <macros>
4 <import>macros.xml</import>
5 <import>cpt-macros.xml</import>
6 </macros>
7 <expand macro="requirements">
8 <requirement type="package" version="0.10.1">gffutils</requirement>
9 </expand>
10 <command detect_errors="aggressive"><![CDATA[
11 python2.7 '$__tool_directory__/searchFile.py'
12 #if $check is not None:
13 --dbaseTerms "$check"
14 #end if
15 #if $term_add.term_selector == "customtxt":
16 --custom_txt "$term_add.custom_txt"
17 #elif $term_add.term_selector == "customfile":
18 --custom_file "$term_add.custom_file"
19 #else:
20 #pass
21 #end if
22 #for $input_select in $input_files:
23 #if $input_select.file_selector.file_select == "gff3selection":
24 --gff3_files #for $gff3_file in $input_select.file_selector.gff3_files:
25 "${gff3_file}" #end for
26 #end if
27 #if $input_select.file_selector.file_select == "gbkselection":
28 --gbk_files #for $gbk_file in $input_select.file_selector.gbk_files:
29 "${gbk_file}" #end for
30 #end if
31 #if $input_select.file_selector.file_select == "faselection":
32 --fa_files #for $fa_file in $input_select.file_selector.fa_files:
33 "${fa_file}" #end for
34 #end if
35 #if $input_select.file_selector.file_select == "blastselection":
36 --blast_files #for $blast_file in $input_select.file_selector.blast_files:
37 "${blast_file}" #end for
38 #end if
39 #end for
40 #if $prox:
41 --prox
42 #end if
43 --output '$output'
44 ]]></command>
45 <inputs>
46 <param label="Using Proximity to Lysis Pipeline?" name="prox" type="boolean" truevalue="--prox" falsevalue="" checked="false" help="required GFF3 input"/>
47 <param name="check" type="select" format="text" optional="true" label="Family terms to search" multiple="true" help="Terms avilable to query from the Lysis-family synonym database (see terms in Shared Data/Lysis family Terms)" display="checkboxes">
48 <option value="endolysins" selected="true">Endolysins</option>
49 <option value="holins" selected="true">Holins and Anti-holins</option>
50 <option value="spanins" selected="true">Spanins</option>
51 <option value="endolysin_domains" selected="true">Endolysin Associated Domains</option>
52 <option value="spanin_domains" selected="true">Spanin Associated Domains</option>
53 <option value="holin_domains" selected="true">Holin Associated Domains</option>
54 </param>
55 <conditional name="term_add">
56 <param name="term_selector" type="select" label="Choose if you'd like to add custom terms">
57 <option value="nocustom" selected="false">No Custom Terms</option>
58 <option value="customtxt" selected="false">Custom Text</option>
59 <option value="customfile" selected="false">Custom File</option>
60 </param>
61 <when value="customtxt">
62 <param name="custom_txt" label="Custom Text" optional="true" type="text" area="true" help="Custom text box for search terms, must be separated by newline (enter)"/>
63 </when>
64 <when value="customfile">
65 <param name="custom_file" label="Custom File" optional="true" type="data" format="txt" help="Custom search terms, uploaded via file, where terms must be separated by newline"/>
66 </when>
67 </conditional>
68 <repeat name="input_files" title="Input Files">
69 <conditional name="file_selector">
70 <param name="file_select" type="select" label="Choose the type of file(s) you'd like to query">
71 <option value="gff3selection" selected="false">GFF3</option>
72 <option value="gbkselection" selected="false">Genbank</option>
73 <option value="faselection" selected="false">FASTA</option>
74 <option value="blastselection" selected="false">BLAST-XML</option>
75 </param>
76 <when value="gff3selection">
77 <param name="gff3_files" label="GFF3 Input" optional="true" multiple="true" type="data" format="gff3"/>
78 </when>
79 <when value="gbkselection">
80 <param name="gbk_files" label="Genbank Input" optional="true" multiple="true" type="data" format="genbank"/>
81 </when>
82 <when value="faselection">
83 <param name="fa_files" label="FASTA Input" optional="true" multiple="true" type="data" format="fasta"/>
84 </when>
85 <when value="blastselection">
86 <param name="blast_files" label="BLAST-xml Input" optional="true" multiple="true" type="data" format="xml"/>
87 </when>
88 </conditional>
89 </repeat>
90 </inputs>
91 <outputs>
92 <data format="txt" name="output" label="termHits">
93 <change_format>
94 <when input="prox" value="--prox" format="gff3"/>
95 </change_format>
96 </data>
97 </outputs>
98 <tests>
99 <test>
100 <param name="prox" value="--prox"/>
101 <param name="check" value="spanins"/>
102 <repeat name="input_files">
103 <conditional name="file_selector">
104 <param name="file_select" value="gff3selection"/>
105 <param name="gff3_files" value="lambda_spaninBLAST.gff3"/>
106 </conditional>
107 </repeat>
108 <output name="output" file="termHits"/>
109 </test>
110 </tests>
111 <help><![CDATA[
112 INPUT : A user's file(s) of interest (gff3, gbk, fasta, or blast-xml) and query said file(s) with a determined set of search terms. The search terms can be either from the curated lysis synonym database, and/or a custom input via file or text box. Additionally, if a user is performing this script in the proximity to lysis workflow, the selector needs to be selected to true, and only gff3 files will be able to complete the job.
113
114 OUTPUT : "termHits.txt" output file has file-type separated returns for features, or lines, that contained hits to a query term. See file search areas, for the regions of each type of file that are searched. If used for the proximity to lysis workflow, the output will be a proxHits.gff3 file that will be able to be fed to the next stages of the pipeline.
115
116 DBase Family Term Descriptions :
117
118 - Endolysins: enzymes that attack one of the structural bonds of the peptidoglycan. The endolysin step requires the hole-forming function of the holins.
119
120 - Holins and Anti-holins: small cytoplasmic membrane proteins that control the timing of lysis by forming a lethal membrane hole at a programmed time (programmed into the holin itself).
121
122 - Spanins: A periplasm-spanning protein complex that disrupts the outer membrane of the host in phage lysis. These complexes are currently identified as either a unimolecular spanin (u-spanin) or two-component system (i-spanin and o-spanin pair).
123
124 - Endolysins Domains: information pulled from PMID: 30873139
125
126 - Spanin Domains: information pulled from http://www.ebi.ac.uk/interpro/search/text/spanin/
127
128 - Holin Domains: information pulled from http://www.ebi.ac.uk/interpro/search/text/holin/
129
130 Additionally, some terms were added by querying QuickGO, located at https://www.ebi.ac.uk/QuickGO/
131
132 File Search Areas:
133
134 - gff3 : entire row
135
136 - genbank : product and note qualifier of each feature
137
138 - fasta : header field
139
140 - blast-xml : hit description
141 ]]></help>
142 <citations>
143 <citation type="doi">10.1371/journal.pcbi.1008214</citation>
144 <citation type="bibtex">
145 @unpublished{galaxyTools,
146 author = {C. Ross},
147 title = {CPT Galaxy Tools},
148 year = {2020-},
149 note = {https://github.com/tamu-cpt/galaxy-tools/}
150 }
151 </citation>
152 </citations>
153 </tool>