Mercurial > repos > bioit_sciensano > phagetermvirome

comparison phage_term_virome.xml @ 0:69e8f12c8b31 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"planemo upload"
author bioit_sciensano
date Fri, 11 Mar 2022 15:06:20 +0000
parents
children ee73cdf35532
comparison
equal deleted inserted replaced
-1:000000000000 0:69e8f12c8b31
1 <tool id="phage_term_virome" name="Phage Term Virome" version="2.0.0" python_template_version="3.5" profile="19.05">
2 <description> Determine phage genome termini and genome packaging mode on single phage or multiple contigs at once. </description>
3 <requirements>
4 <requirement type="package" version="1.0">backports</requirement>
5 <requirement type="package" version="1.6.4">backports.functools_lru_cache</requirement>
6 <requirement type="package" version="0.5">backports_abc</requirement>
7 <requirement type="package" version="0.11.0">cycler</requirement>
8 <requirement type="package" version="1.2.2">libwebp-base</requirement>
9 <requirement type="package" version="1.9.3">lz4-c</requirement>
10 <requirement type="package" version="3.5.1">matplotlib-base</requirement>
11 <requirement type="package" version="3.5.1">matplotlib</requirement>
12 <requirement type="package" version="1.21.5">numpy</requirement>
13 <requirement type="package" version="3.0.0">openssl</requirement>
14 <requirement type="package" version="1.3.5">pandas</requirement>
15 <requirement type="package" version="0.5.2">patsy</requirement>
16 <requirement type="package" version="9.0.1">pillow</requirement>
17 <requirement type="package" version="3.0.7">pyparsing</requirement>
18 <requirement type="package" version="2.8.2">python-dateutil</requirement>
19 <requirement type="package" version="3.7">python_abi</requirement>
20 <requirement type="package" version="2021.3">pytz</requirement>
21 <requirement type="package" version="8.1">readline</requirement>
22 <requirement type="package" version="3.5.68">reportlab</requirement>
23 <requirement type="package" version="1.0.2">scikit-learn</requirement>
24 <requirement type="package" version="1.7.3">scipy</requirement>
25 <requirement type="package" version="59.8.0">setuptools</requirement>
26 <requirement type="package" version="3.6.1">singledispatch</requirement>
27 <requirement type="package" version="0.13.2">statsmodels</requirement>
28 <requirement type="package" version="8.6.12">tk</requirement>
29 <requirement type="package" version="6.1">tornado</requirement>
30 </requirements>
31 <command detect_errors="exit_code"><![CDATA[
32 python3 '$__tool_directory__/PhageTerm.py' -c \${GALAXY_SLOTS} -r $reference --report_title $output_prefix
33
34 ## Manage single-paired data switch
35 #if $single_paired_inputs.input_selector == "single"
36 -f "$single_paired_inputs.fastq"
37 #else
38 #if $single_paired_inputs.input_selector == "paired"
39 -f "$single_paired_inputs.fastq_fw" -p "$single_paired_inputs.fastq_rv"
40 #else
41 -f "$single_paired_inputs.fastq_inputs.forward" -p "$single_paired_inputs.fastq_inputs.reverse"
42 #end if
43 #end if
44
45 #if $seed
46 -s $seed
47 #end if
48
49 #if $surrounding
50 -d $surrounding
51 #end if
52
53 #if $host
54 -g $host
55 #end if
56
57 #if $coverage
58 -m $coverage
59 #end if
60
61 #if $min_phage_lg
62 -l $min_phage_lg
63 #end if
64 ]]></command>
65 <inputs>
66 <conditional name="single_paired_inputs">
67 <param name="input_selector" type="select" label="Single or Paired-end reads" help="Select between paired and single end data">
68 <option value="paired">Paired-end</option>
69 <option value="single">Single-end</option>
70 <option value="paired_collection">Paired-end collection</option>
71 </param>
72 <when value="paired">
73 <param name="fastq_fw" type="data" format="fast" label="Select first set of reads" help="Specify dataset with forward reads"/>
74 <param name="fastq_rv" type="data" format="fastq" label="Select second set of reads" help="Specify dataset with reverse reads"/>
75 </when>
76 <when value="single">
77 <param name="fastq" type="data" format="fastq" label="Select fastq dataset" help="Specify dataset with single reads"/>
78 </when>
79 <when value="paired_collection">
80 <param name="fastq_inputs" format="fastq" type="data_collection" collection_type="paired" label="Select paired-end dataset collection" help="Specify paired dataset collection containing paired reads"/>
81 </when>
82 </conditional>
83 <param name="fastq" format="fastq" type="data" label="Phage reads (FASTQ)" help="Phage sequencing reads file in fastq format from randomly fragmented NGS data (e.g. Illumina TruSeq) but NOT Nextera."/>
84 <param name="reference" format="fasta" type="data" label="Phage genome (FASTA)" help="Phage reference genome file in fasta format, Multi-fasta NOT accepted."/>
85 <param name="output_prefix" type="text" size="20" optional="true" label="Output files prefix" value="Phage" help="Prefix for the output file name."/>
86 <param name="host" format="fasta" type="data" optional="true" label="Bacterial host genome (FASTA)" help="Host reference genome file in fasta format, Multi-fasta NOT accepted. Warning : increases process time."/>
87 <param name="seed" type="integer" optional="true" value="20" label="Seed length" help="Seed length value for alignment of reads."/>
88 <param name="surrounding" type="integer" optional="true" value="20" label="Peak surrounding region" help="Length of the surrounding region defining close peaks to be merged in the analysis process."/>
89 <param name="coverage" type="integer" optional="true" value="250" label="Limit coverage" help="Phage upper limit coverage."/>
90 <param name="min_phage_lg" type="integer" optional="true" value="500" label="Limit phage lenght" help="Minimum phage fasta length."/>
91 </inputs>
92 <outputs>
93 <data name="Statistics" format="csv" label="${output_prefix}_statistics.csv" from_work_dir="*_statistics.csv" />
94 <data name="CohesiveSequence" format="fasta" label="${output_prefix}_cohesive-sequence.fasta" from_work_dir="*_cohesive-sequence.fasta" />
95 <data name="PhageSequence" format="fasta" label="${output_prefix}_sequence.fasta" from_work_dir="*_sequence.fasta" />
96 <data name="PDFReport" format="pdf" label="${output_prefix}_PhageTerm_report.pdf" from_work_dir="*_PhageTerm_report.pdf" />
97 <data name="DirectTermRepeats" format="fasta" label="${output_prefix}_direct-term-repeats.fasta" from_work_dir="*_direct-term-repeats.fasta" />
98 </outputs>
99 <tests>
100 <test>
101 <param name="output_prefix" value="Test-cohesive-5" />
102 <param name="input_selector" value="single" />
103 <param name="fastq" value="COS-5.fastq" />
104 <param name="reference" value="COS-5.fasta" />
105 <output name="Statistics" ftype="csv">
106 <assert_contents>
107 <has_size value="150731" delta="1000" />
108 </assert_contents>
109 </output>
110 <output name="CohesiveSequence" ftype="fasta">
111 <assert_contents>
112 <has_size value="108" delta="50" />
113 </assert_contents>
114 </output>
115 <output name="PhageSequence" ftype="fasta">
116 <assert_contents>
117 <has_size value="3089" delta="100" />
118 </assert_contents>
119 </output>
120 <output name="PDFReport" ftype="pdf">
121 <assert_contents>
122 <has_size value="396227" delta="5000" />
123 </assert_contents>
124 </output>
125 </test>
126 <test>
127 <param name="output_prefix" value="Test-cohesive-3" />
128 <param name="input_selector" value="single" />
129 <param name="fastq" value="COS-3.fastq" />
130 <param name="reference" value="COS-3.fasta" />
131 <output name="Statistics" ftype="csv">
132 <assert_contents>
133 <has_size value="154427" delta="1000" />
134 </assert_contents>
135 </output>
136 <output name="CohesiveSequence" ftype="fasta">
137 <assert_contents>
138 <has_size value="106" delta="50" />
139 </assert_contents>
140 </output>
141 <output name="PhageSequence" ftype="fasta">
142 <assert_contents>
143 <has_size value="3091" delta="100" />
144 </assert_contents>
145 </output>
146 <output name="PDFReport" ftype="pdf">
147 <assert_contents>
148 <has_size value="400459" delta="5000" />
149 </assert_contents>
150 </output>
151 </test>
152 <test>
153 <param name="output_prefix" value="Test-short-direct-terminal-repeats" />
154 <param name="input_selector" value="single" />
155 <param name="fastq" value="DTR-short.fastq" />
156 <param name="reference" value="DTR-short.fasta" />
157 <output name="Statistics" ftype="csv">
158 <assert_contents>
159 <has_size value="165289" delta="1000" />
160 </assert_contents>
161 </output>
162 <output name="DirectTermRepeats" ftype="fasta">
163 <assert_contents>
164 <has_size value="438" delta="50" />
165 </assert_contents>
166 </output>
167 <output name="PhageSequence" ftype="fasta">
168 <assert_contents>
169 <has_size value="3402" delta="100" />
170 </assert_contents>
171 </output>
172 <output name="PDFReport" ftype="pdf">
173 <assert_contents>
174 <has_size value="289115" delta="5000" />
175 </assert_contents>
176 </output>
177 </test>
178 <test>
179 <param name="output_prefix" value="Test-long-direct-terminal-repeats" />
180 <param name="input_selector" value="single" />
181 <param name="fastq" value="DTR-long.fastq" />
182 <param name="reference" value="DTR-long.fasta" />
183 <output name="Statistics" ftype="csv">
184 <assert_contents>
185 <has_size value="1800253" delta="10000" />
186 </assert_contents>
187 </output>
188 <output name="DirectTermRepeats" ftype="fasta">
189 <assert_contents>
190 <has_size value="8615" delta="100" />
191 </assert_contents>
192 </output>
193 <output name="PhageSequence" ftype="fasta">
194 <assert_contents>
195 <has_size value="16759" delta="100" />
196 </assert_contents>
197 </output>
198 <output name="PDFReport" ftype="pdf">
199 <assert_contents>
200 <has_size value="360967" delta="5000" />
201 </assert_contents>
202 </output>
203 </test>
204 <test>
205 <param name="output_prefix" value="Test-Headfull" />
206 <param name="input_selector" value="single" />
207 <param name="fastq" value="Headful.fastq" />
208 <param name="reference" value="Headful.fasta" />
209 <param name="surrounding" value="0" />
210 <output name="Statistics" ftype="csv">
211 <assert_contents>
212 <has_size value="148887" delta="1000" />
213 </assert_contents>
214 </output>
215 <output name="PhageSequence" ftype="fasta">
216 <assert_contents>
217 <has_size value="3078" delta="100" />
218 </assert_contents>
219 </output>
220 <output name="PDFReport" ftype="pdf">
221 <assert_contents>
222 <has_size value="277156" delta="5000" />
223 </assert_contents>
224 </output>
225 </test>
226 <!-- Mu like test absent for two reasons : To hight running time (3 minutes with 10 cores) and crash
227 in case of Mu paired reads because the tool's code try to sum two ranges. Error in 'testMu' : _modules/common_readsCoverage_processing.py", line 385.
228 -->
229 </tests>
230 <help><![CDATA[
231
232 **What it does**
233
234 PhageTermVirome software is a tool to determine phage genome termini and genome packaging mode on single phage or multiple contigs at once.
235 The software uses phage and virome sequencing reads obtained from libraries prepared with DNA fragmented randomly (e.g. Covaris fragmentation,
236 and library preparation using Illumina TruSeq). Phage or virome sequencing reads (fastq files) are aligned to the assembled phage genome or assembled
237 virome (fasta or multifasta files) in order to calculate two types of coverage values (whole genome coverage and the Starting Position Coverage (SPC)). The starting position coverage is used to perform a detailed termini and packaging mode analysis.
238
239 Mu-type phage analysis : can be done if user suspect the phage genome to be Mu-like type (Only for single phage genome analysis, not possible with multifasta file) :
240 User can also provide the host (bacterial) genome sequence. The Mu-type phage analysis will take the reads that does not match the phage
241 genome and align them on the bacterial genome using the same mapping function. The analysis to identify Mu-like phages is available only when providing a single phage genome (not possible if user provide a multi-fast file with multiple assembled phage contigs).
242
243 **Inputs**
244
245 - Raw reads file(s) in fastq format : paired-ends or single-ends.
246 - Phage genome in fasta format
247 - Host genome in fasta format (optional)
248
249 **Outputs**
250
251 - PDF report
252
253 - Statistical table (csv)
254
255 - FASTA file(s) re-organized to start at the predicted termini
256
257 **Tool version**
258
259 *Version:* 4.0.0
260 *Commit SHA:* 5e92822b3f289a329d18914b2183159642abdba4
261
262 ]]></help>
263 <citations>
264 <citation type="doi">10.1101/108100</citation>
265 </citations>
266 </tool>