Mercurial > repos > richard-burhans > segalign

comparison segalign.xml @ 14:f9e91add973c draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/richard-burhans/galaxytools/tree/main/tools/segalign commit eb6ee047d39760f2596b6e343e2dc3898fafc5cd
author richard-burhans
date Mon, 29 Jul 2024 20:53:43 +0000
parents e6fde4082c76
children 99ee8d6f6815
comparison
equal deleted inserted replaced
13:ae2cd39594eb 14:f9e91add973c
6 <import>scoring_options.xml</import> 6 <import>scoring_options.xml</import>
7 <import>seeding_options.xml</import> 7 <import>seeding_options.xml</import>
8 <import>ungapped_extension_options.xml</import> 8 <import>ungapped_extension_options.xml</import>
9 <import>gapped_extension_options.xml</import> 9 <import>gapped_extension_options.xml</import>
10 <import>output_options.xml</import> 10 <import>output_options.xml</import>
11 <import>segalign_output_options.xml</import>
12 <import>system_options.xml</import> 11 <import>system_options.xml</import>
13 </macros> 12 </macros>
14 <expand macro="edam_ontology"/> 13 <expand macro="edam_ontology"/>
15 <expand macro="requirements"/> 14 <expand macro="requirements"/>
16 <required_files> 15 <required_files>
18 <include path="lastz-cmd.ini"/> 17 <include path="lastz-cmd.ini"/>
19 <include path="package_output.py"/> 18 <include path="package_output.py"/>
20 <include path="runner.py"/> 19 <include path="runner.py"/>
21 </required_files> 20 </required_files>
22 <command detect_errors="exit_code"><![CDATA[ 21 <command detect_errors="exit_code"><![CDATA[
23 ## 22 ## Convert input to 2bit for lastz -------------------------------------
24 ## https://www.gnu.org/software/coreutils/manual/html_node/nproc-invocation.html
25 ##
26 ## If the OMP_NUM_THREADS or OMP_THREAD_LIMIT environment variables
27 ## are set, then they will determine the minimum and maximum returned
28 ## value respectively.
29 ##
30 ## This is how you tame nproc(1)
31 ##
32 OMP_THREAD_LIMIT=\${GALAXY_SLOTS:-2}
33 23
34 ## Mode ---------------------------------------------------------------- 24 mkdir -p "\$(pwd)/work" &&
25 faToTwoBit <(gzip -cdfq '$target') "\$(pwd)/work/ref.2bit" &&
26 faToTwoBit <(gzip -cdfq '$query') "\$(pwd)/work/query.2bit" &&
35 27
36 #if str($mode.mode_selector) == "segalign" 28 ## explicitly calling python to bypass a pulsar bug
37 #if str($mode.gapped_extension_options.nogapped) == "false" 29 ## https://github.com/galaxyproject/pulsar/issues/341
38 mkdir -p "\$(pwd)/work" && 30 python '$__tool_directory__/runner.py'
39 faToTwoBit <(gzip -cdfq '$mode.target') "\$(pwd)/work/ref.2bit" && 31 --output-type tarball
40 faToTwoBit <(gzip -cdfq '$mode.query') "\$(pwd)/work/query.2bit" && 32 --output-file '$segalign_output'
41 #end if 33 --diagonal-partition
42 #if str($mode.diagonal_partition_options.diagonal_partition) == "true" 34 --num-cpu \${GALAXY_SLOTS:-2}
43 #set $segalign_mode = "segalign_diagonal_partition" 35 --tool_directory '$__tool_directory__'
44 ## explicitly calling python to bypass a pulsar bug 36 '$target'
45 ## https://github.com/galaxyproject/pulsar/issues/341 37 '$query'
46 python '$__tool_directory__/runner.py'
47 --output-type tarball
48 --output-file '$segalign_diagonal_partition_output'
49 --diagonal-partition
50 --num-cpu \${GALAXY_SLOTS:-2}
51 --tool_directory '$__tool_directory__'
52 #else
53 #set $segalign_mode = "segalign"
54 python '$__tool_directory__/runner.py'
55 --output-type output
56 --output-file '$segalign_output'
57 --num-cpu \${GALAXY_SLOTS:-2}
58 --tool_directory '$__tool_directory__'
59 #end if
60 '$mode.target'
61 '$mode.query'
62 #else if str($mode.mode_selector) == "segalign_repeat_masker"
63 #set $segalign_mode = "segalign_repeat_masker"
64 run_segalign_repeat_masker
65 '$mode.seq_file'
66 --num_cpu \${GALAXY_SLOTS:-2}
67 #end if
68 38
69 ## Sequence Options ---------------------------------------------------- 39 ## Sequence Options ----------------------------------------------------
70 40
71 --strand '$mode.sequence_options.strand_selector' 41 --strand '$sequence_options.strand_selector'
72 #if $segalign_mode == "segalign_repeat_masker"
73 --neighbor_proportion '$mode.sequence_options.neighbor_proportion'
74 #end if
75 42
76 ## Scoring Options ----------------------------------------------------- 43 ## Scoring Options -----------------------------------------------------
77 44
78 #set $scoring_pathname = str($mode.scoring_options.scoring) 45 #set $scoring_pathname = str($scoring_options.scoring)
79 #if $scoring_pathname != "None": 46 #if $scoring_pathname != "None":
80 --scoring '$scoring_pathname' 47 --scoring '$scoring_pathname'
81 #end if 48 #end if
82 #if str($mode.scoring_options.ambiguous_selector) != "x" 49 #if str($scoring_options.ambiguous_selector) != "x"
83 #if str($mode.scoring_options.set_ambiguous_params_selector) == "true" 50 #if str($scoring_options.set_ambiguous_params_selector) == "true"
84 #set $argument_value = ','.join($mode.scoring_options.ambiguous_selector, $mode.scoring_options.ambiguous_reward, $mode.scoring_options.ambiguous_penalty) 51 #set $argument_value = ','.join($scoring_options.ambiguous_selector, $scoring_options.ambiguous_reward, $scoring_options.ambiguous_penalty)
85 --ambiguous '$argument_value' 52 --ambiguous '$argument_value'
86 #else 53 #else
87 --ambiguous '$ambiguous_selector' 54 --ambiguous '$ambiguous_selector'
88 #end if 55 #end if
89 #end if 56 #end if
90 57
91 ## Seeding Options ----------------------------------------------------- 58 ## Seeding Options -----------------------------------------------------
92 59
93 #if str($mode.seeding_options.seed.seed_selector) == "custom" 60 #if str($seeding_options.seed.seed_selector) == "custom"
94 --seed '$mode.seeding_options.seed.custom_seed' 61 --seed '$seeding_options.seed.custom_seed'
95 #else 62 #else
96 --seed '$mode.seeding_options.seed.seed_selector' 63 --seed '$seeding_options.seed.seed_selector'
97 #end if 64 #end if
98 --step '$mode.seeding_options.step' 65 --step '$seeding_options.step'
99 #if str($mode.seeding_options.notransition) == "true" 66 #if str($seeding_options.notransition) == "true"
100 --notransition 67 --notransition
101 #end if 68 #end if
102 69
103 ## Ungapped Extension Options ------------------------------------------ 70 ## Ungapped Extension Options ------------------------------------------
104 71
105 --xdrop '$mode.ungapped_extension_options.xdrop' 72 --xdrop '$ungapped_extension_options.xdrop'
106 --hspthresh '$mode.ungapped_extension_options.hspthresh' 73 --hspthresh '$ungapped_extension_options.hspthresh'
107 #if str($mode.ungapped_extension_options.noentropy) == "true" 74 #if str($ungapped_extension_options.noentropy) == "true"
108 --noentropy 75 --noentropy
109 #end if 76 #end if
110 77
111 ## Gapped Extension Options -------------------------------------------- 78 ## Gapped Extension Options --------------------------------------------
112 79
113 #if $segalign_mode != "segalign_repeat_masker" 80 --nogapped
114 #if str($mode.gapped_extension_options.nogapped) == "true" 81 --ydrop '$gapped_extension_options.ydrop'
115 --nogapped 82 #if str($gapped_extension_options.gappedthresh) != ""
116 #end if 83 --gappedthresh '$gapped_extension_options.gappedthresh'
117 --ydrop '$mode.gapped_extension_options.ydrop' 84 #end if
118 #if str($mode.gapped_extension_options.gappedthresh) != "" 85 #if str($gapped_extension_options.notrivial) == "true"
119 --gappedthresh '$mode.gapped_extension_options.gappedthresh' 86 --notrivial
120 #end if
121 #if str($mode.gapped_extension_options.notrivial) == "true"
122 --notrivial
123 #end if
124 #end if 87 #end if
125 88
126 ## Output Options ----------------------------------------------------- 89 ## Output Options -----------------------------------------------------
127 90
128 #if $segalign_mode != "segalign_repeat_masker" 91 #if str($output_options.format.format_selector) == "bam"
129 #if str($mode.output_options.format.format_selector) == "bam" 92 --format '$output_options.format.bam_options'
130 --format '$mode.output_options.format.bam_options' 93 #else if str($output_options.format.format_selector) == "general_def"
131 #else if str($mode.output_options.format.format_selector) == "general_def" 94 --format general-
132 --format general- 95 #else if str($output_options.format.format_selector) == "maf"
133 #else if str($mode.output_options.format.format_selector) == "general_full" 96 --format '$output_options.format.maf_type'
134 --format 'general-:${mode.output_options.format.fields}' 97 #else if str($output_options.format.format_selector) == "blastn"
135 #else if str($mode.output_options.format.format_selector) == "maf" 98 --format=BLASTN-
136 --format '$mode.output_options.format.maf_type' 99 #else if str($output_options.format.format_selector) == "general_full"
137 #else if str($mode.output_options.format.format_selector) == "blastn" 100 --format 'general-:${output_options.format.fields}'
138 --format=BLASTN- 101 #else if str($output_options.format.format_selector) == "differences"
139 #else if str($mode.output_options.format.format_selector) == "differences" 102 --format=differences
140 --format=differences 103 #end if
141 #end if
142 ## todo :: rplot, bam 104 ## todo :: rplot, bam
143 ## --action:target=multiple 105 ## --action:target=multiple
144 ## $output_format.rplot 106 ## $output_format.rplot
145 ## .if str( $output_format.out.format ) == "bam": 107 ## #if str( $output_format.out.format ) == "bam":
146 ## | samtools sort -@\${GALAXY_SLOTS:-2} -T "\${TMPDIR:-.}" -O bam -o '${output}' 108 ## | samtools sort -@\${GALAXY_SLOTS:-2} -T "\${TMPDIR:-.}" -O bam -o '${output}'
147 ## .else: 109 ## #else:
148 ## > '${output}' 110 ## > '${output}'
149 ## .end if 111 ## #end if
150 ## .if $output_format.rplot: 112 ## #if $output_format.rplot:
151 ## && 113 ## &&
152 ## Rscript $r_plot > /dev/null 2>&1 114 ## Rscript $r_plot > /dev/null 2>&1
153 ## .end if 115 ## #end if
154 #else if $segalign_mode == "segalign_repeat_masker"
155 --M '$mode.output_options.M'
156 --output '$segalign_repeat_masker_output'
157 #end if
158 #if str($mode.output_options.markend) == "true"
159 --markend
160 #end if
161 116
162 ## System Options ----------------------------------------------------- 117 ## System Options -----------------------------------------------------
163 118
164 --wga_chunk_size '$mode.system_options.wga_chunk_size' 119 --wga_chunk_size '$system_options.wga_chunk_size'
165 --lastz_interval_size '$mode.system_options.lastz_interval_size' 120 --lastz_interval_size '$system_options.lastz_interval_size'
166 --seq_block_size '$mode.system_options.seq_block_size' 121 --seq_block_size '$system_options.seq_block_size'
167 --num_gpu '$mode.system_options.num_gpu' 122 --num_gpu '$system_options.num_gpu'
168 #if str($mode.system_options.debug) == "true" 123 #if str($system_options.debug) == "true"
169 --debug 124 --debug
170 #end if 125 #end if
171 126
172 ## ------------------------------------------------------------------- 127 ## Package Output ----------------------------------------------------
173 128
174 #if $segalign_mode == "segalign_diagonal_partition" 129 && python '$__tool_directory__/package_output.py'
175 &&
176 python '$__tool_directory__/package_output.py'
177 --tool_directory '$__tool_directory__'
178 #end if
179 130
180 ]]></command> 131 ]]></command>
181 <inputs> 132 <inputs>
182 <conditional name="mode"> 133 <param name="target" type="data" format="fasta,fasta.gz" label="Target sequence file in FASTA format"/>
183 <param name="mode_selector" type="select" label="Choose the mode"> 134 <param name="query" type="data" format="fasta,fasta.gz" label="Query sequence file in FASTA format"/>
184 <option value="segalign" selected="true">SegAlign</option> 135 <expand macro="sequence_options"/>
185 <option value="segalign_repeat_masker">SegAlign repeat masker</option> 136 <expand macro="scoring_options"/>
186 </param> 137 <expand macro="seeding_options"/>
187 <when value="segalign"> 138 <expand macro="ungapped_extension_options"/>
188 <param name="target" type="data" format="fasta,fasta.gz" label="Target sequence file in FASTA format"/> 139 <expand macro="gapped_extension_options"/>
189 <param name="query" type="data" format="fasta,fasta.gz" label="Query sequence file in FASTA format"/> 140 <expand macro="output_options"/>
190 <expand macro="sequence_options"/> 141 <expand macro="system_options"/>
191 <expand macro="scoring_options"/>
192 <expand macro="seeding_options"/>
193 <expand macro="ungapped_extension_options"/>
194 <expand macro="gapped_extension_options"/>
195 <expand macro="output_options">
196 <expand macro="segalign_output_options"/>
197 </expand>
198 <expand macro="system_options"/>
199 <section name="diagonal_partition_options" expanded="false" title="Diagonal Partition Options">
200 <param argument="--diagonal_partition" type="boolean" value="false" label="Enable diagonal partition optimization"/>
201 </section>
202 </when>
203 <when value="segalign_repeat_masker">
204 <param name="seq_file" type="data" format="fasta,fasta.gz" label="Sequence file in FASTA format"/>
205 <expand macro="sequence_options">
206 <param argument="--neighbor_proportion" type="float" value="0.2" label="Proportion of neighbouring intervals to align the query interval to"/>
207 </expand>
208 <expand macro="scoring_options"/>
209 <expand macro="seeding_options"/>
210 <expand macro="ungapped_extension_options"/>
211 <expand macro="output_options">
212 <param argument="--M" type="integer" value="1" max="255" label="report any position that is covered by at least this many alignments; the maximum allowed depth is 255"/>
213 </expand>
214 <expand macro="system_options"/>
215 </when>
216 </conditional>
217 </inputs> 142 </inputs>
218 <outputs> 143 <outputs>
219 <data name="segalign_output" format="tabular" label="SegAlign on ${on_string}"> 144 <data name="segalign_output" format="tgz" from_work_dir="data_package.tgz" label="SegAlign on ${on_string}"/>
220 <change_format>
221 <when input="mode.output_options.format.format_selector" value="bam" format="bam"/>
222 <when input="mode.output_options.format.format_selector" value="maf" format="maf"/>
223 <when input="mode.output_options.format.format_selector" value="differences" format="interval"/>
224 </change_format>
225 <filter>mode['mode_selector'] == 'segalign' and mode['diagonal_partition_options']['diagonal_partition'] is False</filter>
226 </data>
227 <data name="segalign_diagonal_partition_output" format="tgz" from_work_dir="data_package.tgz" label="SegAlign Diagonal Partition on ${on_string}">
228 <filter>mode['mode_selector'] == 'segalign' and mode['diagonal_partition_options']['diagonal_partition'] is True</filter>
229 </data>
230 <data name="segalign_repeat_masker_output" format="tabular" label="SegAlign Repeat Masker on ${on_string}">
231 <filter>mode['mode_selector'] == 'segalign_repeat_masker'</filter>
232 </data>
233 </outputs> 145 </outputs>
234 <tests> 146 <tests>
235 <test expect_num_outputs="1" expect_test_failure="true"> 147 <test expect_num_outputs="1" expect_test_failure="true">
236 <param name="mode_selector" value="segalign"/>
237 <param name="target" value="hg38.chr20.chunk.fa.gz" ftype="fasta.gz"/> 148 <param name="target" value="hg38.chr20.chunk.fa.gz" ftype="fasta.gz"/>
238 <param name="query" value="mm39.chr2.chunk.fa.gz" ftype="fasta.gz"/> 149 <param name="query" value="mm39.chr2.chunk.fa.gz" ftype="fasta.gz"/>
239 <output name="segalign_output" decompress="true" file="segalign-output.maf.gz" ftype="maf"/> 150 <output name="segalign_output" ftype="tgz">
240 </test>
241 <test expect_num_outputs="1" expect_test_failure="true">
242 <param name="mode_selector" value="segalign_repeat_masker"/>
243 <param name="seq_file" value="hg38.chr20.chunk.fa.gz" ftype="fasta.gz"/>
244 <output name="segalign_repeat_masker_output" decompress="true" file="segalign-repeat-masker-output.tab.gz" ftype="tabular"/>
245 </test>
246 <test expect_num_outputs="1" expect_test_failure="true">
247 <param name="mode_selector" value="segalign"/>
248 <param name="target" value="hg38.chr20.chunk.fa.gz" ftype="fasta.gz"/>
249 <param name="query" value="mm39.chr2.chunk.fa.gz" ftype="fasta.gz"/>
250 <param name="diagonal_partition" value="true"/>
251 <output name="segalign_diagonal_partition_output" ftype="tgz">
252 <assert_contents> 151 <assert_contents>
253 <has_archive_member path="galaxy/commands.json"/> 152 <has_archive_member path="galaxy/commands.json"/>
254 </assert_contents> 153 </assert_contents>
255 </output> 154 </output>
256 </test> 155 </test>