Mercurial > repos > nml > spades

comparison spades.xml @ 7:9006e5836729 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/spades commit cde78013f2df8cc06372d73f6332f7ab1f120a25
author iuc
date Mon, 30 Oct 2017 07:20:13 -0400
parents 65c2d63fcbe6
children 884dc0264950
comparison
equal deleted inserted replaced
6:65c2d63fcbe6 7:9006e5836729
1 <tool id="spades" name="SPAdes" version="3.9.0"> 1 <tool id="spades" name="SPAdes" version="3.11.1">
2 <description>genome assembler for regular and single-cell projects</description> 2 <description>genome assembler for regular and single-cell projects</description>
3 <requirements> 3 <requirements>
4 <requirement type="package" version="3.9.0">spades</requirement> 4 <requirement type="package" version="3.11.1">spades</requirement>
5 </requirements> 5 </requirements>
6 <stdio> 6 <stdio>
7 <exit_code range="1:" /> 7 <exit_code range="1:" />
8 </stdio> 8 </stdio>
9 <command> 9 <command>
72 #end for 72 #end for
73 && cat contigs.fasta | python '$write_tsv_script' > '$out_contig_stats' 73 && cat contigs.fasta | python '$write_tsv_script' > '$out_contig_stats'
74 && cat scaffolds.fasta | python '$write_tsv_script' > '$out_scaffold_stats' 74 && cat scaffolds.fasta | python '$write_tsv_script' > '$out_scaffold_stats'
75 ]]> 75 ]]>
76 </command> 76 </command>
77
78 <configfiles>
79 <configfile name="write_tsv_script"><![CDATA[#!/usr/bin/env python
80 import sys,re
81 search_str = r'^>(NODE|\S+)_(\d+)(?:_|\s)length_(\d+)_cov_(\d+\.*\d*).*\$'
82 replace_str = r'\1_\2\t\3\t\4'
83 cmd = re.compile(search_str)
84 sys.stdout.write('#name\tlength\tcoverage\n')
85 for i,line in enumerate(sys.stdin):
86 if cmd.match(line):
87 sys.stdout.write(cmd.sub(replace_str,line))
88 ]]>
89 </configfile>
90 </configfiles>
91
77 <inputs> 92 <inputs>
78 <param argument="--sc" falsevalue="" help="This option is required for MDA (single-cell) data." label="Single-cell?" name="sc" truevalue="--sc" type="boolean"> 93 <param argument="--sc" falsevalue="" help="This option is required for MDA (single-cell) data." label="Single-cell?" name="sc" truevalue="--sc" type="boolean">
79 <option value="false">No</option> 94 <option value="false">No</option>
80 <option value="true">Yes</option> 95 <option value="true">Yes</option>
81 </param> 96 </param>
140 <param optional="true" format="fastq" label="PacBio CLR reads" multiple="true" name="pacbio_reads" type="data" /> 155 <param optional="true" format="fastq" label="PacBio CLR reads" multiple="true" name="pacbio_reads" type="data" />
141 <param optional="true" format="fastq" label="Nanopore reads" multiple="true" name="nanopore_reads" type="data" /> 156 <param optional="true" format="fastq" label="Nanopore reads" multiple="true" name="nanopore_reads" type="data" />
142 <param optional="true" format="fasta,fastq" label="Sanger reads" multiple="true" name="sanger_reads" type="data" /> 157 <param optional="true" format="fasta,fastq" label="Sanger reads" multiple="true" name="sanger_reads" type="data" />
143 <param optional="true" format="fasta,fastq" label="Trusted contigs" multiple="true" name="trusted_contigs" type="data" /> 158 <param optional="true" format="fasta,fastq" label="Trusted contigs" multiple="true" name="trusted_contigs" type="data" />
144 <param optional="true" format="fasta,fastq" label="Untrusted contigs" multiple="true" name="untrusted_contigs" type="data" /> 159 <param optional="true" format="fasta,fastq" label="Untrusted contigs" multiple="true" name="untrusted_contigs" type="data" />
160 <param name="contig_graph_out" type="boolean" checked="False" label="Output final assembly graph (contigs)?" help="Will output the final assembly graph (contigs) in fastg format for visualisation" />
161 <param name="scaffold_graph_out" type="boolean" checked="False" label="Output final assembly graph with scaffolds?" help="Will output the final assembly graph with scaffold information in gfa format for visualisation" />
145 </inputs> 162 </inputs>
146 <configfiles>
147 <configfile name="write_tsv_script"><![CDATA[#!/usr/bin/env python
148 import sys,re
149 search_str = r'^>(NODE|\S+)_(\d+)(?:_|\s)length_(\d+)_cov_(\d+\.*\d*).*\$'
150 replace_str = r'\1_\2\t\3\t\4'
151 cmd = re.compile(search_str)
152 sys.stdout.write('#name\tlength\tcoverage\n')
153 for i,line in enumerate(sys.stdin):
154 if cmd.match(line):
155 sys.stdout.write(cmd.sub(replace_str,line))
156 ]]>
157 </configfile>
158 </configfiles>
159 163
160 <outputs> 164 <outputs>
161 <data format="tabular" label="SPAdes contig stats" name="out_contig_stats" > 165 <data format="tabular" label="${tool.name} on ${on_string}: contig stats" name="out_contig_stats" >
162 <actions> 166 <actions>
163 <action name="column_names" type="metadata" default="name,length,coverage"/> 167 <action name="column_names" type="metadata" default="name,length,coverage"/>
164 </actions> 168 </actions>
165 </data> 169 </data>
166 <data format="tabular" label="SPAdes scaffold stats" name="out_scaffold_stats" > 170 <data format="tabular" label="${tool.name} on ${on_string}: scaffold stats" name="out_scaffold_stats" >
167 <actions> 171 <actions>
168 <action name="column_names" type="metadata" default="name,length,coverage"/> 172 <action name="column_names" type="metadata" default="name,length,coverage"/>
169 </actions> 173 </actions>
170 </data> 174 </data>
171 <data format="fasta" from_work_dir="contigs.fasta" label="SPAdes contigs (fasta)" name="out_contigs" /> 175 <data format="fasta" from_work_dir="contigs.fasta" label="${tool.name} on ${on_string}: contigs (fasta)" name="out_contigs" />
172 <data format="fasta" from_work_dir="scaffolds.fasta" label="SPAdes scaffolds (fasta)" name="out_scaffolds" /> 176 <data format="fasta" from_work_dir="scaffolds.fasta" label="${tool.name} on ${on_string}: scaffolds (fasta)" name="out_scaffolds" />
173 <data format="txt" from_work_dir="spades.log" label="SPAdes log" name="out_log" /> 177 <data format="txt" from_work_dir="spades.log" label="${tool.name} on ${on_string}: log" name="out_log" />
178 <data format="txt" from_work_dir="assembly_graph.fastg" label="${tool.name} on ${on_string}: assembly graph" name="contig_graph">
179 <filter>contig_graph_out</filter>
180 </data>
181 <data format="txt" from_work_dir="assembly_graph_with_scaffolds.gfa" label="${tool.name} on ${on_string}: assembly graph with scaffolds" name="scaffold_graph">
182 <filter>scaffold_graph_out</filter>
183 </data>
174 </outputs> 184 </outputs>
175 <tests> 185 <tests>
176 <test> 186 <test> <!-- Test 1 - basic test with k=33 -->
177 <param name="sc" value="false" /> 187 <param name="sc" value="false" />
178 <param name="careful" value="false" /> 188 <param name="careful" value="false" />
179 <param name="kmers" value="33" /> 189 <param name="kmers" value="33" />
180 <param name="lib_type" value="paired_end" /> 190 <param name="lib_type" value="paired_end" />
181 <param ftype="fastq" name="fwd_reads" value="ecoli_1K_1.fq" /> 191 <param ftype="fastq" name="fwd_reads" value="ecoli_1K_1.fq" />
182 <param ftype="fastq" name="rev_reads" value="ecoli_1K_2.fq" /> 192 <param ftype="fastq" name="rev_reads" value="ecoli_1K_2.fq" />
183 <output compare="re_match" file="kmer_33_output.fa" ftype="fasta" lines_diff="1" name="out_contigs" /> 193 <output compare="re_match" file="kmer_33_output.fa" ftype="fasta" lines_diff="1" name="out_contigs" />
184 <output name="out_contig_stats"> 194 <output name="out_contig_stats">
185 <assert_contents> 195 <assert_contents>
186 <has_text_matching expression="NODE_1\t1000\t225"/> 196 <has_text_matching expression="NODE_1\t1000"/>
187 </assert_contents> 197 </assert_contents>
188 </output> 198 </output>
189 </test> 199 </test>
190 <test> 200 <test> <!-- Test 2 - auto k -->
191 <param name="sc" value="false" /> 201 <param name="sc" value="false" />
192 <param name="careful" value="false" /> 202 <param name="careful" value="false" />
193 <param name="auto_kmer_choice" value="true" /> 203 <param name="auto_kmer_choice" value="true" />
194 <param name="lib_type" value="paired_end" /> 204 <param name="lib_type" value="paired_end" />
195 <param ftype="fastq" name="fwd_reads" value="ecoli_1K_1.fq" /> 205 <param ftype="fastq" name="fwd_reads" value="ecoli_1K_1.fq" />
196 <param ftype="fastq" name="rev_reads" value="ecoli_1K_2.fq" /> 206 <param ftype="fastq" name="rev_reads" value="ecoli_1K_2.fq" />
197 <output compare="re_match" file="auto_kmer_output.fa" ftype="fasta" lines_diff="1" name="out_contigs" /> 207 <output compare="re_match" file="auto_kmer_output.fa" ftype="fasta" lines_diff="1" name="out_contigs" />
198 </test> 208 </test>
199 <test> 209 <test> <!-- Test 3 - k=77 -->
200 <param name="sc" value="false" /> 210 <param name="sc" value="false" />
201 <param name="careful" value="false" /> 211 <param name="careful" value="false" />
202 <param name="kmers" value="77" /> 212 <param name="kmers" value="77" />
203 <param name="lib_type" value="paired_end" /> 213 <param name="lib_type" value="paired_end" />
204 <param ftype="fastq" name="fwd_reads" value="ecoli_1K_1.fq" /> 214 <param ftype="fastq" name="fwd_reads" value="ecoli_1K_1.fq" />
205 <param ftype="fastq" name="rev_reads" value="ecoli_1K_2.fq" /> 215 <param ftype="fastq" name="rev_reads" value="ecoli_1K_2.fq" />
206 <output compare="re_match" file="kmer_77_output.fa" ftype="fasta" lines_diff="1" name="out_contigs" /> 216 <output compare="re_match" file="kmer_77_output.fa" ftype="fasta" lines_diff="1" name="out_contigs" />
217 </test>
218 <test> <!-- Test 4 - test for extra graph outputs -->
219 <param name="sc" value="false" />
220 <param name="careful" value="false" />
221 <param name="kmers" value="33" />
222 <param name="lib_type" value="paired_end" />
223 <param ftype="fastq" name="fwd_reads" value="ecoli_1K_1.fq" />
224 <param ftype="fastq" name="rev_reads" value="ecoli_1K_2.fq" />
225 <param name="contig_graph_out" value="true" />
226 <param name="scaffold_graph_out" value="true" />
227 <output compare="re_match" file="kmer_33_output.fa" ftype="fasta" lines_diff="1" name="out_contigs" />
228 <output name="out_contig_stats">
229 <assert_contents>
230 <has_text_matching expression="NODE_1\t1000"/>
231 </assert_contents>
232 </output>
233 <output name="contig_graph">
234 <assert_contents>
235 <has_text text=">EDGE_"/>
236 </assert_contents>
237 </output>
238 <output name="scaffold_graph">
239 <assert_contents>
240 <has_text text="NODE_"/>
241 </assert_contents>
242 </output>
207 </test> 243 </test>
208 </tests> 244 </tests>
209 <help> 245 <help>
210 <![CDATA[ 246 <![CDATA[
211 **What it does** 247 **What it does**
212 248
213 SPAdes – St. Petersburg genome assembler – is intended for both standard isolates and single-cell MDA bacteria assemblies. See http://bioinf.spbau.ru/en/spades for more details on SPAdes. 249 SPAdes – St. Petersburg genome assembler – is intended for both standard isolates and single-cell MDA bacteria assemblies. See http://bioinf.spbau.ru/en/spades for more details on SPAdes.
214 250
215 This wrapper runs SPAdes 3.9, collects the output, and throws away all the temporary files. It also produces a tab file with contig names, length and coverage. 251 This wrapper runs SPAdes, collects the output, and throws away all the temporary files. It also produces a tab file with contig names, length and coverage.
216 252
217 **License** 253 **License**
218 254
219 SPAdes is developed by and copyrighted to Saint-Petersburg Academic University, and is released under GPLv2. 255 SPAdes is developed by and copyrighted to Saint-Petersburg Academic University, and is released under GPLv2.
220 256
229 Original wrapper developed by Lionel Guy. 265 Original wrapper developed by Lionel Guy.
230 266
231 Anton Korobeynikov greatlty helped understanding how SPAdes work, and integrated handy features into SPAdes. 267 Anton Korobeynikov greatlty helped understanding how SPAdes work, and integrated handy features into SPAdes.
232 268
233 Nicola Soranzo fixed various bugs. 269 Nicola Soranzo fixed various bugs.
270
271 Simon Gladman added fastg optional outputs.
234 ]]> 272 ]]>
235 </help> 273 </help>
236 <citations> 274 <citations>
237 <citation type="doi">10.1089/cmb.2012.0021</citation> 275 <citation type="doi">10.1089/cmb.2012.0021</citation>
238 </citations> 276 </citations>