Mercurial > repos > edward-kirton > roche454_toolsuite

comparison runMapping.xml @ 3:bf1f8bc4abe6 default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
minor updates for v2.6
author eskirton@lbl.gov
date Wed, 21 Dec 2011 19:42:53 -0800
parents 2d86d5b112e8
children
comparison
equal deleted inserted replaced
2:2d86d5b112e8 3:bf1f8bc4abe6
1 <tool id="runMapping" name="runMapping" version="1.0.0"> 1 <tool id="runMapping" name="runMapping" version="1.0.1">
2 <description>Map Roche/454 reads to a reference using Newbler</description> 2 <description>Map Roche/454 reads to a reference using Newbler</description>
3 <command interpreter='perl'>runMapping_wrapper.pl 3 <command interpreter='perl'>runMapping_wrapper.pl
4 $newbler_metrics.extra_files_path 4 $newbler_metrics.extra_files_path
5 $alignment_info 5 $alignment_info
6 $all_contigs_fasta 6 $all_contigs_fasta
17 $tag_pair_align 17 $tag_pair_align
18 $trim_status 18 $trim_status
19 $trimmed_reads_fasta 19 $trimmed_reads_fasta
20 $trimmed_reads_qual 20 $trimmed_reads_qual
21 $contigs_ace 21 $contigs_ace
22 $contigs_bam
22 $large_contigs_fasta 23 $large_contigs_fasta
23 $large_contigs_qual 24 $large_contigs_qual
24 $gene_status 25 $gene_status
25 $newbler_exe -o $newbler_metrics.extra_files_path 26 runAssembly
27 -o $newbler_metrics.extra_files_path
26 -cpu 8 28 -cpu 8
27 -a $a 29 -a $a
28 -e $e 30 -e $e
29 -mi $mi 31 -mi $mi
30 -ml $ml 32 -ml $ml
32 $pair 34 $pair
33 $info 35 $info
34 $notrim 36 $notrim
35 $tr 37 $tr
36 $ace 38 $ace
39 $bam
37 $no 40 $no
38 $qo 41 $qo
39 $nor 42 $nor
40 $ud 43 $ud
41 -ss $ss 44 -ss $ss
68 #for $i in $sanger_inputs 71 #for $i in $sanger_inputs
69 ${i.sanger_input} 72 ${i.sanger_input}
70 #end for 73 #end for
71 </command> 74 </command>
72 <inputs> 75 <inputs>
73 <!-- NEWBLER VERSION -->
74 <param name='newbler_exe' type='select' display='radio' label='Newbler version'>
75 <option value='/jgi/tools/454/rig-DataProcessing_2.3/bin/runMapping'>2.3</option>
76 <option value='/jgi/tools/454/rig-DataProcessing_2.4pre-20091204/bin/runMapping'>2.4</option>
77 <option value='/home/copeland/local/x86_64/newbler/v2.5p1-internal-10Jun23-1/runMapping' selected='true'>2.5</option>
78 </param>
79
80 <!-- READSEQ INFILES --> 76 <!-- READSEQ INFILES -->
81 <repeat name="sff_inputs" title="Unpaired Reads Sff Files"> 77 <repeat name="sff_inputs" title="Unpaired Reads Sff Files">
82 <param name="sff_input" type="data" format="sff" label="SE Sff file"/> 78 <param name="sff_input" type="data" format="sff" label="SE Sff file"/>
83 </repeat> 79 </repeat>
84 <repeat name="sanger_inputs" title="Unpaired Reads Fasta Files"> 80 <repeat name="sanger_inputs" title="Unpaired Reads Fasta/Fastq Files">
85 <param name="sanger_input" type="data" format="fasta" label="SE Fasta file"/> 81 <param name="sanger_input" type="data" format="fasta,fastqsanger" label="SE Fasta/Fastq file"/>
86 </repeat> 82 </repeat>
87 <repeat name="sff_paired_inputs" title="Paired Reads Sff Files"> 83 <repeat name="sff_paired_inputs" title="Paired Reads Sff Files">
88 <param name="sff_paired_input" type="data" format="sff" label="PE Sff file"/> 84 <param name="sff_paired_input" type="data" format="sff" label="PE Sff file"/>
89 </repeat> 85 </repeat>
90 <repeat name="sanger_paired_inputs" title="Paired Reads Fasta Files"> 86 <repeat name="sanger_paired_inputs" title="Paired Reads Fasta/Fastq Files">
91 <param name="sanger_paired_input" type="data" format="fasta" label="PE Fasta file"/> 87 <param name="sanger_paired_input" type="data" format="fasta,fastqsanger" label="PE Fasta/Fastq file"/>
92 </repeat> 88 </repeat>
93 <param name='paired_reads' type='select' display='radio' label='[-paired_reads] If supplying paired reads (above), do you want paired-read info?'> 89 <param name="paired_reads" type="select" display="radio" label="[-paired_reads] If supplying paired reads (above), do you want paired-read info?">
94 <option value='false'>no</option> 90 <option value="false">no</option>
95 <option value='true'>[-paired_reads] yes</option> 91 <option value="true">[-paired_reads] yes</option>
96 </param> 92 </param>
97 <param name='pair' type='select' display='radio' label='[-pair] Output pairwise overlaps'> 93 <param name="pair" type="select" display="radio" label="[-pair] Output pairwise overlaps">
98 <option value=''>no</option> 94 <option value="">no</option>
99 <option value='-pair'>[-pair] yes</option> 95 <option value="-pair">[-pair] yes</option>
100 </param> 96 </param>
101 97
102 <!-- SOURCE DNA TYPE --> 98 <!-- SOURCE DNA TYPE -->
103 <param name='l' type="integer" value='500' label="[-l] This option sets the minimum length for a contig to appear in the 454LargeContigs.fna file"/> 99 <param name="l" type="integer" value="500" label="[-l] This option sets the minimum length for a contig to appear in the 454LargeContigs.fna file"/>
104 100
105 <!-- INPUTS SPECIFIC TO MAPPING --> 101 <!-- INPUTS SPECIFIC TO MAPPING -->
106 <repeat name='ref_inputs' title='Reference Sequence'> 102 <repeat name="ref_inputs" title="Reference Sequence">
107 <param name='ref_input' type='data' format='fasta' label='Fasta file'/> 103 <param name="ref_input" type="data" format="fasta" label="Fasta file"/>
108 </repeat> 104 </repeat>
109 <param name='ref_type' type='select' display='radio' label='Reference type'> 105 <param name="ref_type" type="select" display="radio" label="Reference type">
110 <option value='-gref'>[-gref] Genomic reference sequence</option> 106 <option value="-gref">[-gref] Genomic reference sequence</option>
111 <option value='-cref'>[-cref] cDNA reference sequence</option> 107 <option value="-cref">[-cref] cDNA reference sequence</option>
112 </param> 108 </param>
113 <param name='rst' type='integer' value='12' label='[-rst] Repeat score threshold parameter. Allowed values: 0 or greater'/> 109 <param name="rst" type="integer" value="12" label="[-rst] Repeat score threshold parameter. Allowed values: 0 or greater"/>
114 <param name='hsl' type='integer' value='70' label='[-hsl] Hit-per-seed limit parameter'/> 110 <param name="hsl" type="integer" value="70" label="[-hsl] Hit-per-seed limit parameter"/>
115 <param name='srv' type='boolean' truevalue='-srv' falsevalue='' checked='false' label='[-srv] Single read variant output'/> 111 <param name="srv" type="boolean" truevalue="-srv" falsevalue="" checked="false" label="[-srv] Single read variant output"/>
116 112
117 <!-- OPTIONAL ARGUMENTS --> 113 <!-- OPTIONAL ARGUMENTS -->
118 <!-- NYI 114 <!-- NYI
119 <param name='accno' type='data' format='tabular' optional='true' label='[-accno] Specify annotation data. Required only if reference sequence headers do not contain gene=NAME pairs' /> 115 <param name="accno" type="data" format="tabular" optional="true" label="[-accno] Specify annotation data. Required only if reference sequence headers do not contain gene=NAME pairs" />
120 <param name='annot' type='data' format='tabular' optional='true' label='[-annot] Supply gene, transcript, and protein information' /> 116 <param name="annot" type="data" format="tabular" optional="true" label="[-annot] Supply gene, transcript, and protein information" />
121 --> 117 -->
122 <param name='mcf' type='data' format='tabular' optional='true' label='[-mcf] Specify non-default MID config file' /> 118 <param name="mcf" type="data" format="tabular" optional="true" label="[-mcf] Specify non-default MID config file" />
123 <param name='fi' type='data' format='txt' optional='true' label='[-fi] Include filter file to be specified' /> 119 <param name="fi" type="data" format="txt" optional="true" label="[-fi] Include filter file to be specified" />
124 <param name='fe' type='data' format='txt' optional='true' label='[-fe] Exclude filter file to be specified' /> 120 <param name="fe" type="data" format="txt" optional="true" label="[-fe] Exclude filter file to be specified" />
125 <param name='vt' type='data' format='fasta' optional='true' label="[-vt] This option specifies a vector trimming database, or FASTA file of sequences to be used to trim the ends of input reads (for cloning vectors, primers, adapters or other end sequences)" /> 121 <param name="vt" type="data" format="fasta" optional="true" label="[-vt] This option specifies a vector trimming database, or FASTA file of sequences to be used to trim the ends of input reads (for cloning vectors, primers, adapters or other end sequences)" />
126 <param name='vs' type='data' format='fasta' optional='true' label="[-vs] This option specifies a vector screening database, or FASTA file of sequences to be used to screen the input reads for contaminants. Reads that completely align against the screening database are trimmed completely (so that it is not used in the computation), but otherwise the read trimpoints are not changed" /> 122 <param name="vs" type="data" format="fasta" optional="true" label="[-vs] This option specifies a vector screening database, or FASTA file of sequences to be used to screen the input reads for contaminants. Reads that completely align against the screening database are trimmed completely (so that it is not used in the computation), but otherwise the read trimpoints are not changed" />
127 123
128 124
129 <!-- READ TRIMMING --> 125 <!-- READ TRIMMING -->
130 <param name='minlen' type='integer' value='20' label='[-minlen] Minimum length of reads to use (15-45 allowed)'/> 126 <param name="minlen" type="integer" value="20" label="[-minlen] Minimum length of reads to use (15-45 allowed)"/>
131 <param name='notrim' type='boolean' truevalue='-notrim' falsevalue='' checked='false' label='[-notrim] Do not perform default quality and primer trimming of input reads'/> 127 <param name="notrim" type="boolean" truevalue="-notrim" falsevalue="" checked="false" label="[-notrim] Do not perform default quality and primer trimming of input reads"/>
132 <param name='tr' type='select' display='radio' label='[-tr] Output trimmed reads'> 128 <param name="tr" type="select" display="radio" label="[-tr] Output trimmed reads">
133 <option value=''>no</option> 129 <option value="">no</option>
134 <option value='-tr'>[-tr] yes</option> 130 <option value="-tr">[-tr] yes</option>
135 </param> 131 </param>
136 <param name='nor' type='boolean' truevalue='-nor' falsevalue='' label='[-nor] Turn off the automatic rescore function for read quality scores'/> 132 <param name="nor" type="boolean" truevalue="-nor" falsevalue="" label="[-nor] Turn off the automatic rescore function for read quality scores"/>
137 <param name='ud' type='boolean' truevalue='-ud' falsevalue='' label='[-ud] Treat each read separately, with no grouping of duplicates'/> 133 <param name="ud" type="boolean" truevalue="-ud" falsevalue="" label="[-ud] Treat each read separately, with no grouping of duplicates"/>
138 134
139 <!-- ALIGNMENT PARAMETERS --> 135 <!-- ALIGNMENT PARAMETERS -->
140 <param name='ss' type='integer' value='12' label='[-ss] Seed step parameter - The number of bases between seed generation locations used in the exact k-mer matching part of the overlap detection. Allow values: 1 or greater'/> 136 <param name="ss" type="integer" value="12" label="[-ss] Seed step parameter - The number of bases between seed generation locations used in the exact k-mer matching part of the overlap detection. Allow values: 1 or greater"/>
141 <param name='sl' type='integer' value='16' label='[-sl] Seed length parameter - The number of bases used for each seed in the exact k-mer matching part of the overlap detection (i.e. the "k" value of the k-mer matching). Allowed values: 6-16'/> 137 <param name="sl" type="integer" value="16" label="[-sl] Seed length parameter - The number of bases used for each seed in the exact k-mer matching part of the overlap detection (i.e. the 'k' value of the k-mer matching). Allowed values: 6-16"/>
142 <param name='sc' type='integer' value='1' label='[-sc] Seed count parameter - The number of seeds required in a window before an extension is made. Allowed values: 1 or greater'/> 138 <param name="sc" type="integer" value="1" label="[-sc] Seed count parameter - The number of seeds required in a window before an extension is made. Allowed values: 1 or greater"/>
143 <param name='ml' type="text" value='40' label="[-ml] Minimum overlap length - The minimum length of overlaps used for the pairwise alignment step. The value can either be a minimum length in bases or a percentage of read length. In the case of a percentage, simply include '%' immediately following the numeric value. Allowed values: 1 or greater"/> 139 <param name="ml" type="text" value="40" label="[-ml] Minimum overlap length - The minimum length of overlaps used for the pairwise alignment step. The value can either be a minimum length in bases or a percentage of read length. In the case of a percentage, simply include '%' immediately following the numeric value. Allowed values: 1 or greater"/>
144 <param name='mi' type="integer" value='90' label="[-mi] Minimum overlap identity - The percent identity of overlaps used for the pairwise alignment step. Allowed values: 0 or greater"/> 140 <param name="mi" type="integer" value="90" label="[-mi] Minimum overlap identity - The percent identity of overlaps used for the pairwise alignment step. Allowed values: 0 or greater"/>
145 <param name='ais' type='integer' value='2' label='[-ais] Alignment identity score - When multiple overlaps are found, the per-overlap column identity score used to sort the overlaps for use in the progressive alignment. Allowed values: 0 or greater'/> 141 <param name="ais" type="integer" value="2" label="[-ais] Alignment identity score - When multiple overlaps are found, the per-overlap column identity score used to sort the overlaps for use in the progressive alignment. Allowed values: 0 or greater"/>
146 142
147 <!-- ASSEMBLY OPTIONS --> 143 <!-- ASSEMBLY OPTIONS -->
148 <param name='e' type="integer" value='0' label="[-e] This option tells the assembler that the expected depth of the data is at a certain level. The assembler has been optimized for datasets in the 10-50x oversampling size, and this option helps the assembler with datasets that have a higher oversampling level. A value of 0 resets the assembler computation to use its default algorithms"/> 144 <param name="e" type="integer" value="0" label="[-e] This option tells the assembler that the expected depth of the data is at a certain level. The assembler has been optimized for datasets in the 10-50x oversampling size, and this option helps the assembler with datasets that have a higher oversampling level. A value of 0 resets the assembler computation to use its default algorithms"/>
149 145
150 <!-- OUTPUT OPTIONS --> 146 <!-- OUTPUT OPTIONS -->
151 <param name='no' type='select' display='radio' label='[-no] Do complete assembly'> 147 <param name="no" type="select" display="radio" label="[-no] Do complete assembly">
152 <option value=''>do complete assembly</option> 148 <option value="">do complete assembly</option>
153 <option value='-no'>[-no] do not assemble; do alignments only</option> 149 <option value="-no">[-no] do not assemble; do alignments only</option>
154 </param> 150 </param>
155 <param name='qo' type='boolean' truevalue='' falsevalue='-qo' checked='false' label='[-qo] Generate quick output for mapping and assembly. Disables signal distribution computation for calling consensus sequences and can decrease accuracy'/> 151 <param name="qo" type="boolean" truevalue="" falsevalue="-qo" checked="false" label="[-qo] Generate quick output for mapping and assembly. Disables signal distribution computation for calling consensus sequences and can decrease accuracy"/>
156 <param name='a' type="integer" value='100' label="[-a] This option sets the minimum length for a contig to appear in the 454AllContigs.fna file."/> 152 <param name="a" type="integer" value="100" label="[-a] This option sets the minimum length for a contig to appear in the 454AllContigs.fna file."/>
157 <param name='info' type='select' display='radio' label='Output Alignment Info'> 153 <param name="info" type="select" display="radio" label="Output Alignment Info">
158 <option value='-info'>[-info] yes</option> 154 <option value="-info">[-info] yes</option>
159 <option value='-infoall'>[-infoall] yes, including 0-coverage positions</option> 155 <option value="-infoall">[-infoall] yes, including 0-coverage positions</option>
160 </param> 156 </param>
161 <param name='ace' type='select' display='radio' label='Produce Ace assembly file'> 157 <param name="ace" type="select" display="radio" label="Produce Ace assembly file">
162 <option value=''>no</option> 158 <option value="">no</option>
163 <option value='-ace'>[-ace] yes</option> 159 <option value="-ace">[-ace] yes</option>
164 </param> 160 </param>
161 <param name="bam" type="boolean" truevalue="-bam" falsevalue="-nobam" checked="true" label="[-bam] Produce Bam alignments file" />
165 </inputs> 162 </inputs>
166 163
167 <outputs> 164 <outputs>
168 <!-- the following are common to runMapping and runAssembly --> 165 <!-- the following are common to runMapping and runAssembly -->
169 <data name='newbler_metrics' format='txt' /> 166 <data name="newbler_metrics" format="txt" />
170 <data name='read_status' format='tabular' label='Read Status'/> 167 <data name="read_status" format="tabular" label="${tool.name} on $on_string: Read Status"/>
171 <data name='trimmed_reads_fasta' format='fasta' label='Trimmed Reads (Fasta)'> 168 <data name="trimmed_reads_fasta" format="fasta" label="${tool.name} on $on_string: Trimmed Reads (Fasta)">
172 <filter>tr == '-tr'</filter> 169 <filter>tr == "-tr"</filter>
173 </data> 170 </data>
174 <data name='trimmed_reads_qual' format='qual454' label='Trimmed Reads (Qual)'> 171 <data name="trimmed_reads_qual" format="qual454" label="${tool.name} on $on_string: Trimmed Reads (Qual)">
175 <filter>tr == '-tr'</filter> 172 <filter>tr == "-tr"</filter>
176 </data> 173 </data>
177 <!-- the following produced only if no != '-no' --> 174 <!-- the following produced only if no != "-no" -->
178 <data name='alignment_info' format='tabular' label='Alignment Info'/> 175 <data name="alignment_info" format="tabular" label="${tool.name} on $on_string: Alignment Info"/>
179 <data name='all_contigs_fasta' format='fasta' label='All Contigs (Fasta)'> 176 <data name="all_contigs_fasta" format="fasta" label="${tool.name} on $on_string: All Contigs (Fasta)">
180 <filter>no != '-no'</filter> 177 <filter>no != "-no"</filter>
181 </data> 178 </data>
182 <data name='all_contigs_qual' format='qual454' label='All Contigs (Qual454)'> 179 <data name="all_contigs_qual" format="qual454" label="${tool.name} on $on_string: All Contigs (Qual454)">
183 <filter>no != '-no'</filter> 180 <filter>no != "-no"</filter>
184 </data> 181 </data>
185 <data name='contigs_ace' format='ace' label='Contigs (Ace)'> 182 <data name="contigs_ace" format="ace" label="${tool.name} on $on_string: Contigs (Ace)">
186 <filter>ace == '-ace' and no != '-no'</filter> 183 <filter>ace == "-ace" and no != "-no"</filter>
187 </data> 184 </data>
188 <data name='large_contigs_fasta' format='fasta' label='Large Contigs (Fasta)'> 185 <data name="contigs_bam" format="bam" label="${tool.name} on $on_string: Contigs (Bam)">
189 <filter>no != '-no'</filter> 186 <filter>bam is True and no != "-no"</filter>
190 </data> 187 </data>
191 <data name='large_contigs_qual' format='qual454' label='Large Contigs (Qual454)'> 188 <data name="large_contigs_fasta" format="fasta" label="${tool.name} on $on_string: Large Contigs (Fasta)">
192 <filter>no != '-no'</filter> 189 <filter>no != "-no"</filter>
193 </data> 190 </data>
194 <data name='pair_align' format='txt' label='Pairwise Alignments'> 191 <data name="large_contigs_qual" format="qual454" label="${tool.name} on $on_string: Large Contigs (Qual454)">
195 <filter>pair == '-pair' and no != '-no'</filter> 192 <filter>no != "-no"</filter>
196 </data> 193 </data>
197 <data name='pair_status' format='tabular' label='Paired-End Read Status'> 194 <data name="pair_align" format="txt" label="${tool.name} on $on_string: Pairwise Alignments">
198 <filter>paired_reads == 'true' and no != '-no'</filter> 195 <filter>pair == "-pair" and no != "-no"</filter>
199 </data> 196 </data>
200 <data name='scaffolds_fasta' format='fasta' label='Scaffolds (Fasta)'> 197 <data name="pair_status" format="tabular" label="${tool.name} on $on_string: Paired-End Read Status">
201 <filter>paired_reads == 'true' and no != '-no'</filter> 198 <filter>paired_reads == "true" and no != "-no"</filter>
202 </data> 199 </data>
203 <data name='scaffolds_qual' format='qual454' label='Scaffolds (Qual454)'> 200 <data name="scaffolds_fasta" format="fasta" label="${tool.name} on $on_string: Scaffolds (Fasta)">
204 <filter>paired_reads == 'true' and no != '-no'</filter> 201 <filter>paired_reads == "true" and no != "-no"</filter>
205 </data> 202 </data>
206 <data name='scaffolds_agp' format='tabular' label='Scaffolds (Agp)'> 203 <data name="scaffolds_qual" format="qual454" label="${tool.name} on $on_string: Scaffolds (Qual454)">
207 <filter>paired_reads == 'true' and no != '-no'</filter> 204 <filter>paired_reads == "true" and no != "-no"</filter>
208 </data> 205 </data>
209 <data name='tag_pair_align' format='txt' label='Tag Pair Alignments'> 206 <data name="scaffolds_agp" format="tabular" label="${tool.name} on $on_string: Scaffolds (Agp)">
210 <filter>pair == '-pair' and paired_reads == 'true' and no != '-no'</filter> 207 <filter>paired_reads == "true" and no != "-no"</filter>
211 </data> 208 </data>
212 <data name='trim_status' format='tabular' label='Trim Status'/> 209 <data name="tag_pair_align" format="txt" label="${tool.name} on $on_string: Tag Pair Alignments">
210 <filter>pair == "-pair" and paired_reads == "true" and no != "-no"</filter>
211 </data>
212 <data name="trim_status" format="tabular" label="${tool.name} on $on_string: Trim Status"/>
213 213
214 <!-- THE FOLLOWING ARE LIMITED TO MAPPING --> 214 <!-- THE FOLLOWING ARE LIMITED TO MAPPING -->
215 <data name='all_diffs' format='tabular' label='All Diffs'/> 215 <data name="all_diffs" format="tabular" label="${tool.name} on $on_string: All Diffs"/>
216 <data name='all_struct_vars' format='tabular' label='All Struct Vars'/> 216 <data name="all_struct_vars" format="tabular" label="${tool.name} on $on_string: All Struct Vars"/>
217 <data name='hc_diff' format='tabular' label='High Confidence Diff'/> 217 <data name="hc_diff" format="tabular" label="${tool.name} on $on_string: High Confidence Diff"/>
218 <data name='hc_struct_vars' format='tabular' label='High Confidence Struct Vars'/> 218 <data name="hc_struct_vars" format="tabular" label="${tool.name} on $on_string: High Confidence Struct Vars"/>
219 <data name='gene_status' format='tabular' label='Gene Status'/> 219 <data name="gene_status" format="tabular" label="${tool.name} on $on_string: Gene Status"/>
220 <data name='mapping_qc' format='xls' label='Mapping QC (Excel)'/> 220 <data name="mapping_qc" format="xls" label="${tool.name} on $on_string: Mapping QC (Excel)"/>
221 <data name='ref_status' format='tabular' label='Ref Status'/> 221 <data name="ref_status" format="tabular" label="${tool.name} on $on_string: Ref Status"/>
222 222
223 </outputs> 223 </outputs>
224 <help> 224 <help>
225 225
226 **What it does** 226 **What it does**