Mercurial > repos > nml > pneumocat

comparison pneumocat.xml @ 0:33de0245ed99 draft default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"planemo upload for repository https://github.com/phe-bioinformatics/PneumoCaT commit c1002f7ad15e676357c6489878291de07bbde841"
author nml
date Tue, 24 Mar 2020 13:27:46 -0400
parents
children
comparison
equal deleted inserted replaced
-1:000000000000 0:33de0245ed99
1 <tool id="pneumocat" name="PneumoCaT" version="@VERSION@">
2 <description> Pneumococcal Capsular Typing of illumina reads</description>
3 <macros>
4 <import>macros.xml</import>
5 </macros>
6 <expand macro="requirements" />
7 <command detect_errors="exit_code">
8 <![CDATA[
9
10 #import os
11 #import re
12
13 #def check_ending($name, $forward_read=True)
14 ## Pneumocat needs to have name_1.fastq and name_2.fastq to work
15 ## Check for correct ending and change ending if needed
16
17 #if re.search(r'(\.|_)\S*(1|2)*$', $name)
18 #if $forward_read
19 #return re.sub(r'(\.|_)\S*(1|2)*$', '_R1.fastq', $name)
20 #else
21 #return re.sub(r'(\.|_)\S*(1|2)*$', '_R2.fastq', $name)
22 #end if
23
24 #else
25 #if $forward_read
26 #return '{}_R1.fastq'.format($name)
27 #else
28 #return '{}_R2.fastq'.format($name)
29 #end if
30 #end if
31 #end def
32
33
34 #if $input.type == 'paired'
35 #set $initial = re.sub('[^\w_]', '_', os.path.splitext($input.forward.name)[0])
36
37 #set $for_input = $check_ending($initial)
38 #set $rev_input = $check_ending($initial, forward_read=False)
39
40 ln -s '$input.forward' ./$for_input &&
41 ln -s '$input.reverse' ./$rev_input &&
42
43 #elif $input.type == 'paired_collection'
44 #set $initial = re.sub('[^\w_]', '_', os.path.splitext($input.fastq_collection.forward.name)[0])
45
46 #set $for_input = $check_ending($initial)
47 #set $rev_input = $check_ending($initial, forward_read=False)
48
49 ln -s '$input.fastq_collection.forward' ./$for_input &&
50 ln -s '$input.fastq_collection.reverse' ./$rev_input &&
51
52 #end if
53
54 PneumoCaT.py -1 '$for_input' -2 '$rev_input' -o outputs --threads '\${GALAXY_SLOTS:-1}' --cleanup
55 ]]>
56 </command>
57 <inputs>
58 <conditional name="input">
59 <param name="type" type="select" label="Sequence Data Type">
60 <option value="paired">Paired-end reads (FASTQ)</option>
61 <option value="paired_collection">Paired-end reads collection (FASTQ)</option>
62 </param>
63 <when value="paired">
64 <param name="forward"
65 type="data" format="fastq,fastqsanger,fastq.gz,fastqsanger.gz"
66 optional="false"
67 multiple="false"
68 label="Forward reads (FASTQ)"
69 help="Must have ASCII encoded quality scores"
70 />
71 <param name="reverse"
72 type="data" format="fastq,fastqsanger,fastq.gz,fastqsanger.gz"
73 optional="false"
74 label="Reverse reads (FASTQ)"
75 help="File format must match the Forward FASTQ file"
76 />
77 </when>
78 <when value="paired_collection">
79 <param name="fastq_collection"
80 type="data_collection" format="fastq,fastqsanger, fastq.gz, fastqsanger.gz"
81 collection_type="paired"
82 optional="false"
83 label="Paired-end reads collection (FASTQ)"
84 />
85 </when>
86 </conditional>
87 </inputs>
88 <outputs>
89 <data format="txt" name="coverage_summary" from_work_dir="outputs/coverage_summary.txt" label="PneumoCaT Coverage Summary.txt"/>
90 <data format="xml" name="results" from_work_dir="outputs/*.results.xml" label="PneumoCaT Results.xml"/>
91 <data format="xml" name="specific_results" from_work_dir="outputs/SNP_based_serotyping/*.results.xml" label="PneumoCaT Serotype Distinction.xml"/>
92 <data format="txt" name="variant_summary" from_work_dir="outputs/SNP_based_serotyping/variant_summary.yml" label="PneumoCaT Variant Summary.yml"/>
93 </outputs>
94 <tests>
95 <test>
96 <conditional name="input">
97 <param name="type" value="paired" />
98 <param name="forward" value="09N_R1.fastq" />
99 <param name="reverse" value="09N_R2.fastq" />
100 </conditional>
101 <output name="coverage_summary" file="coverage_summary.txt" />
102 <output name="results" file="results.xml" />
103 </test>
104 <test>
105 <conditional name="input">
106 <param name="type" value="paired_collection" />
107 <param name="fastq_collection">
108 <collection type="paired">
109 <element name="forward" value="09N_R1.fastq.gz" ftype="fastq.gz" />
110 <element name="reverse" value="09N_R2.fastq.gz" ftype="fastq.gz" />
111 </collection>
112 </param>
113 </conditional>
114 <output name="coverage_summary" file="coverage_summary.txt" />
115 <output name="results" file="results.xml" />
116 </test>
117 </tests>
118 <help>
119 <![CDATA[
120
121 PneumoCaT
122 ---------
123
124 PneumoCaT (Pneumococcal Capsular Typing) uses a two-step step approach to assign capsular type to
125 S.pneumoniae genomic data (Illumina). More info can be found at the `PneumoCaT github page <https://github.com/phe-bioinformatics/PneumoCaT>`_
126
127 Program Steps
128 #############
129
130 - **Step 1:** Reads from each readset are mapped to capsular locus sequences for all known capsular types using bowtie2
131
132 - This step is considered successful if the readset matches > 90% to one or more capsular locus sequences
133
134 - If only a singular capsular locus is matched, PneumoCaT terminates and reports that as the assigned capsular type
135
136 - If more than 1 loci are matched then the tool moves to step 2
137
138 - **Step 2:** Variant calling with the capsular type variant database
139
140 - Used to distinguish serotypes within a serogroup/genogroup
141
142 Please note PneumoCaT applies a quality metric requiring a mean depth of 20 reads across the mapped sequence
143 and a minimum depth of 5 reads for mapping. The report will retrun "Failed" if these conditions are not met.
144
145 Inputs
146 ######
147
148 - **Paired-end Illumina reads** with one of the following example formats is prefered:
149
150 - <name>_1.fastq and <name>_2.fastq
151
152 - <name>_R1.fastq and <name>_R2.fastq
153
154 - <name>_R1.fastqsanger.gz and <name>_R2.fastqsanger.gz
155
156 - If the reads are not formatted as above, the wrapper will append _R1.fastq and _R2.fastq to allow function
157
158 Outputs
159 #######
160
161 Please see `**PneumoCaTs interpreting results document** <https://github.com/phe-bioinformatics/PneumoCaT/blob/master/Documentation/InterpretingResults.pdf>`_
162 for full up-to-date information on how to interpret PneumoCaT results.
163
164 Galaxy will output 4 results running PneumoCaT with 2 of the results only appearing if **Step 2** variant calling is done
165
166 **1. Coverage Summary.txt** -- Always output unless fails --
167
168 **2. Results.xml** -- Always output unless fails --
169
170 **3. Serotype distinction.xml** -- Step 2 Required --
171
172 **4. Variant Summary.txt** -- Step 2 Required --
173
174
175 **Note** - Galaxy will always output files 3 and 4 even if step 2 is not done. In these cases, the files will have no data.
176
177 ]]>
178 </help>
179 <expand macro="citations" />
180 </tool>