Mercurial > repos > yating-l > ucsc_blat

comparison blat.xml @ 34:53cf9d25ef39 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload commit c5f669e334bf65c1158855bd1eef88df78a8028c
author iuc
date Mon, 21 Nov 2022 11:11:06 +0000
parents f90551569666
children 87644259e668
comparison
equal deleted inserted replaced
33:43803843920b 34:53cf9d25ef39
1 <?xml version="1.0"?> 1 <tool id="ucsc_blat" name="UCSC BLAT Alignment Tool" version="@TOOL_VERSION@+galaxy@VERSION_SUFFIX@">
2 <tool id="ucsc_blat" name="UCSC BLAT Alignment Tool" version="1.1"> 2 <description>BLAST-like sequence alignment tool</description>
3 <description>Standalone blat sequence search command line tool</description> 3 <macros>
4 <token name="@TOOL_VERSION@">377</token>
5 <token name="@VERSION_SUFFIX@">0</token>
6 </macros>
7 <xrefs>
8 <xref type="bio.tools">blat</xref>
9 </xrefs>
4 <requirements> 10 <requirements>
5 <requirement type="package" version="1.0">ucsc_tools_340_for_BLAT</requirement> 11 <requirement type="package" version="@TOOL_VERSION@">ucsc-blat</requirement>
6 </requirements> 12 </requirements>
7 <command detect_errors="exit_code"><![CDATA[ 13 <command detect_errors="exit_code"><![CDATA[
8 #if str($reference_source.reference_source_selector) == "history": 14 #if str($reference_source.reference_source_selector) == "history":
9 #set $reference_fasta_filename = "localref.fa" 15 ## blat depends on file extension
10 ln -s '${reference_source.database}' '${reference_fasta_filename}' && 16 #if $reference_source.database.is_of_type("fasta"):
11 #else: 17 #set $reference_fasta_filename = "localref.fa"
12 #set $reference_fasta_filename = str($reference_source.database.fields.path) 18 #elif $reference_source.database.is_of_type("twobit"):
13 #end if 19 #set $reference_fasta_filename = "localref.2bit"
14 blat 20 #else
15 #if $noHead == "yes" 21 #set $reference_fasta_filename = "localref"
16 -noHead 22 #end if
17 #end if 23 ln -s '$reference_source.database' '$reference_fasta_filename' &&
18 -q=$query_type 24 #else:
19 -t=$database_type 25 #set $reference_fasta_filename = str($reference_source.database.fields.path)
20 #if $mask_type.mask == "file.out": 26 #end if
21 -mask=$mask_type.mask_file 27
22 #else: 28 blat
23 -mask=$mask_type.mask 29 -q=$query_type
24 #end if 30 -t=$database_type
25 '$reference_fasta_filename' 31 $oneOff
26 '${query}' 32 #if str($minScore)
27 output 33 -minScore=$minScore
28 && sort -k 10,10 -k 12,12n output > '${output_sorted}' 34 #end if
29 ]]></command> 35 -maxGap=$maxGap
30 <inputs> 36 #if str($repMatch)
31 <conditional name="reference_source"> 37 -repMatch=$repMatch
32 <param name="reference_source_selector" type="select" label="Choose the source for the database"> 38 #end if
33 <option value="cached">Locally cached</option> 39 #if $mask_type.mask == "file.out":
34 <option value="history">History</option> 40 -mask='$mask_type.mask_file'
35 </param> 41 #else:
36 <when value="cached"> 42 -mask=$mask_type.mask
37 <param name="database" type="select" label="Select database"> 43 #end if
38 <options from_data_table="all_fasta"> 44 #if str($dots)
39 <filter type="sort_by" column="2" /> 45 -dots=$dots
40 </options> 46 #end if
41 <validator type="no_options" message="A built-in database is not available" /> 47 $trimT
42 </param> 48 $noTrimA
43 </when> 49 $trimHardA
44 <when value="history"> 50 $fastMap
45 <param name="database" type="data" format="fasta,twobit" label="Using database file, either a .fa, .nib or .2bit file" /> 51 $fine
46 </when> 52 #if str($maxIntron)
47 </conditional> 53 -maxIntron=$maxIntron
48 <param type="data" name="query" format="fasta,twobit" label="Query data, either a .fa, .nib or .2bit file"/> 54 #end if
49 <param type="select" name="database_type" format="text" multiple="false" label="database type" help="Choose your database type, the default is dnax" argument="-t"> 55 $extendThroughN
50 <option value="dna">dna - DNA sequence</option> 56 '$reference_fasta_filename'
51 <option value="prot">prot - protein sequence</option> 57 '$query'
52 <option value="dnax" selected="true">dnax - DNA sequence translated in six frames to protein</option> 58 -out=$out
59 '$output'
60 ]]></command>
61 <inputs>
62 <conditional name="reference_source">
63 <param name="reference_source_selector" type="select" label="Choose the source for the database">
64 <option value="cached">Locally cached</option>
65 <option value="history">History</option>
53 </param> 66 </param>
54 <param type="select" name="query_type" format="text" multiple="false" label="query type" help="Choose your query type, the default is rnax" argument="-q"> 67 <when value="cached">
55 <option value="dna">dna - DNA sequence </option> 68 <param name="database" type="select" label="Select database">
56 <option value="rna">rna - RNA sequence</option> 69 <options from_data_table="all_fasta">
57 <option value="prot">prot - protein sequence</option> 70 <filter type="sort_by" column="2" />
58 <option value="dnax">dnax - DNA sequence translated in six frames to protein</option> 71 </options>
59 <option value="rnax" selected="true">rnax - DNA sequence translated in three frames to protein</option> 72 <validator type="no_options" message="A built-in database is not available" />
73 </param>
74 </when>
75 <when value="history">
76 <param name="database" type="data" format="fasta, twobit" label="Using database file, either a .fa, .nib or .2bit file" />
77 </when>
78 </conditional>
79 <param name="query" type="data" format="fasta, twobit" label="Query data, either a .fa, .nib or .2bit file"/>
80 <param argument="-t" name="database_type" type="select" format="txt" multiple="false" label="database type" help="Choose your database type, the default is dnax">
81 <option value="dna">dna - DNA sequence</option>
82 <option value="prot">prot - protein sequence</option>
83 <option value="dnax" selected="true">dnax - DNA sequence translated in six frames to protein</option>
84 </param>
85 <param argument="-q" name="query_type" type="select" format="txt" multiple="false" label="query type" help="Choose your query type, the default is rnax">
86 <option value="dna">dna - DNA sequence </option>
87 <option value="rna">rna - RNA sequence</option>
88 <option value="prot">prot - protein sequence</option>
89 <option value="dnax">dnax - DNA sequence translated in six frames to protein</option>
90 <option value="rnax" selected="true">rnax - DNA sequence translated in three frames to protein</option>
91 </param>
92 <param argument="-oneOff" type="boolean" truevalue="-oneOff=1" falsevalue="" label="If set, this allows one mismatch in tile and still triggers an alignments" />
93 <param argument="-minScore" type="integer" value="30" label="Minimum score" help="It is the matches minus the mismatches minus some sort of gap penalty" />
94 <param argument="-maxGap" type="integer" value="2" min="0" max="3" label="Maximum gap between tiles in a clump" help="Usually set from 0 to 3. Only relevant for minMatch > 1" />
95 <param argument="-repMatch" type="integer" value="" optional="true" label="Number of repetitions of a tile allowed before it is marked as overused" help="Typically this is 256 for tileSize 12, 1024 for tileSize 11, 4096 for tileSize 10. Also affected by stepSize. When stepSize is halved repMatch is doubled to compensate" />
96 <conditional name="mask_type">
97 <param argument="-mask" type="select" label="Mask out repeats" help="Alignments won't be started in masked region but may extend through it in nucleotide searches. Masked areas are ignored entirely in protein or translated searches. Default is lower">
98 <option value="lower" selected="true">lower - mask out lower-cased sequence</option>
99 <option value="upper">upper - mask out upper-cased sequence</option>
100 <option value="out">out - mask according to database.out RepeatMasker .out file</option>
101 <option value="file.out">file.out - mask database according to RepeatMasker file.out</option>
60 </param> 102 </param>
61 <param type="select" name="noHead" format="text" label="Suppresses .psl header (so it's just a tab-separated file)." > 103 <when value="lower" />
62 <option value="yes" selected="true">Yes, suppresses .psl header</option> 104 <when value="upper" />
63 <option value="no">No, do not suppresses .psl header</option> 105 <when value="out" />
64 </param> 106 <when value="file.out">
65 <conditional name="mask_type"> 107 <param name="mask_file" type="data" format="txt" label="RepeatMasker file.out" />
66 <param name="mask" type="select" label="Mask out repeats" help="Alignments won't be started in masked region 108 </when>
67 but may extend through it in nucleotide searches. Masked areas 109 </conditional>
68 are ignored entirely in protein or translated searches. Default is lower" argument="-mask"> 110 <param argument="-dots" type="integer" value="" optional="true" label="Output a dot every N sequences in log" help="Dots show program's progress" />
69 <option value="lower" selected="true">lower - mask out lower-cased sequence</option> 111 <param argument="-trimT" type="boolean" truevalue="-trimT" falsevalue="" label="Trim leading poly-T" />
70 <option value="upper">upper - mask out upper-cased sequence</option> 112 <param argument="-noTrimA" type="boolean" truevalue="-noTrimA" falsevalue="" label="Don't trim trailing poly-A" />
71 <option value="out">out - mask according to database.out RepeatMasker .out file</option> 113 <param argument="-trimHardA" type="boolean" truevalue="-trimHardA" falsevalue="" label="Remove poly-A tail from qSize and alignments in .psl output" />
72 <option value="file.out">file.out - mask database according to RepeatMasker file.out</option> 114 <param argument="-fastMap" type="boolean" truevalue="-fastMap" falsevalue="" label="Run for fast DNA/DNA remapping" help="It does not allow introns and require high %ID. Query sizes must not exceed 5000" />
73 </param> 115 <param argument="-fine" type="boolean" truevalue="-fine" falsevalue="" label="Refine search for small initial and terminal exons" help="For high-quality mRNAs. Not recommended for ESTs" />
74 <when value="lower" /> 116 <param argument="-maxIntron" type="integer" value="750000" optional="true" label="Maximum intron size" />
75 <when value="upper" /> 117 <param argument="-extendThroughN" type="boolean" truevalue="-extendThroughN" falsevalue="" label="Allow extension of alignment through large blocks of N's" />
76 <when value="out" /> 118 <param name="out" type="select" label="Select output file format (-out)">
77 <when value="file.out"> 119 <option value="psl">Tab-separated format, no sequence (psl)</option>
78 <param name="mask_file" type="data" format="txt" label="RepeatMasker file.out" /> 120 <option value="psl -noHead">Tab-separated format, no sequence, no header (psl -noHead)</option>
79 </when> 121 <option value="axt">Blastz-associated axt format (axt)</option>
80 </conditional> 122 <option value="maf">Multiz-associated maf format (maf)</option>
81 </inputs> 123 <option value="sim4">Similar to sim4 format (sim4)</option>
82 <outputs> 124 <option value="wublast">Similar to WU-BLAST format (wublast)</option>
83 <data format="psl" name="output_sorted" label="${tool.name} on ${on_string}"></data> 125 <option value="blast">Similar to NCBI BLAST format (blast)</option>
84 </outputs> 126 <option value="blast8">NCBI BLAST tabular format (blast8)</option>
85 <tests> 127 <option value="blast9">NCBI BLAST tabular format with comments (blast9)</option>
86 <!-- test on query of GenBank RefSeq records for Gallus gallus and database of Amazona vittata --> 128 </param>
87 <test> 129 </inputs>
130 <outputs>
131 <data name="output" format="tabular" label="${tool.name} on ${on_string}">
132 <change_format>
133 <when input="out" value="axt" format="axt" />
134 <when input="out" value="maf" format="maf" />
135 <when input="out" value="sim4" format="txt" />
136 <when input="out" value="wublast" format="tabular" />
137 <when input="out" value="blast" format="tabular" />
138 </change_format>
139 </data>
140 </outputs>
141 <tests>
142 <!-- test on query of GenBank RefSeq records for Gallus gallus and database of Amazona vittata -->
143 <test>
88 <param name="reference_source_selector" value="history" /> 144 <param name="reference_source_selector" value="history" />
89 <param name="database" value="amaVit1_Gallus/amaVit1.fa" /> 145 <param name="database" value="amaVit1_Gallus/amaVit1.fa" />
90 <param name="query" value="amaVit1_Gallus/Gallus_gallus_RefSeq.fa" /> 146 <param name="query" value="amaVit1_Gallus/Gallus_gallus_RefSeq.fa" />
91 <param name="database_type" value="dnax" /> 147 <param name="database_type" value="dnax" />
92 <param name="query_type" value="rnax" /> 148 <param name="query_type" value="rnax" />
93 <param name="noHead" value="yes" />
94 <param name="mask" value="lower" /> 149 <param name="mask" value="lower" />
95 <output name="output_sorted" value="amaVit1_Gallus/amaVit1_Gallus_gallus.psl" /> 150 <param name="out" value="psl -noHead" />
96 </test> 151 <output name="output" value="amaVit1_Gallus/amaVit1_Gallus_gallus_sorted.psl" sort="true"/>
97 <!-- test on query of partial mRNA of Drosophila melanogaster and the database of Drosophila biamipes dot chromosome --> 152 </test>
98 <test> 153 <!-- test on query of partial mRNA of Drosophila melanogaster and the database of Drosophila biamipes dot chromosome -->
154 <test>
99 <param name="reference_source_selector" value="history" /> 155 <param name="reference_source_selector" value="history" />
100 <param name="database" value="dbia3/dbia3.fa" /> 156 <param name="database" value="dbia3/dbia3.fa" />
101 <param name="query" value="dbia3/dmel-transcript.fa" /> 157 <param name="query" value="dbia3/dmel-transcript.fa" />
102 <param name="database_type" value="dnax" /> 158 <param name="database_type" value="dnax" />
103 <param name="query_type" value="rnax" /> 159 <param name="query_type" value="rnax" />
104 <param name="noHead" value="yes" />
105 <param name="mask" value="lower" /> 160 <param name="mask" value="lower" />
106 <output name="output_sorted" value="dbia3/dbia3.sorted.psl" /> 161 <param name="out" value="psl -noHead" />
107 </test> 162 <param name="maxIntron" value="" />
108 <!-- test on the database masked by repeat masker --> 163 <output name="output" value="dbia3/dbia3.sorted.psl" sort="true"/>
109 <test> 164 </test>
165 <!-- test on the database masked by repeat masker -->
166 <test>
110 <param name="reference_source_selector" value="history" /> 167 <param name="reference_source_selector" value="history" />
111 <param name="database" value="dbia3/dbia3_masked.2bit" /> 168 <param name="database" value="dbia3/dbia3_masked.2bit" />
112 <param name="query" value="dbia3/dmel-transcript.fa" /> 169 <param name="query" value="dbia3/dmel-transcript.fa" />
113 <param name="database_type" value="dnax" /> 170 <param name="database_type" value="dnax" />
114 <param name="query_type" value="rnax" /> 171 <param name="query_type" value="rnax" />
115 <param name="noHead" value="yes" /> 172 <param name="oneOff" value="false" />
173 <param name="minScore" value="30" />
174 <param name="maxGap" value="2" />
175 <param name="trimT" value="false" />
176 <param name="noTrimA" value="false" />
177 <param name="fine" value="false" />
178 <param name="maxIntron" value="750000" />
179 <param name="extendThroughN" value="false" />
116 <param name="mask" value="file.out" /> 180 <param name="mask" value="file.out" />
117 <param name="mask_file" value="dbia3/dbia3_RM.out" /> 181 <param name="mask_file" value="dbia3/dbia3_RM.out" />
118 <output name="output_sorted" value="dbia3/dbia3_masked.sorted.psl" /> 182 <param name="out" value="psl -noHead" />
119 </test> 183 <output name="output" value="dbia3/dbia3_masked.sorted.psl"/>
120 184 </test>
121 </tests> 185 </tests>
122 <help> 186 <help>
123 <![CDATA[ 187 <![CDATA[
124 BLAT 188 BLAT
125 ==== 189 ====
126 BLAT is a bioinformatics software a tool which performs rapid mRNA/DNA and cross-species protein alignments. 190 BLAT is a bioinformatics software a tool which performs rapid mRNA/DNA and cross-species protein alignments.
127 191
128 blat (version: v340)- Standalone blat sequence search command line tool. 192 blat (version: v36)- Standalone blat sequence search command line tool.
129 ------------------------------------------------------------------------- 193 -------------------------------------------------------------------------
130 194
131 usage: 195 usage:
132 ++++++ 196 ++++++
133 197
137 database and query are each either a .fa, .nib or .2bit file, 201 database and query are each either a .fa, .nib or .2bit file,
138 or a list of these files with one file name per line. 202 or a list of these files with one file name per line.
139 -ooc=11.ooc tells the program to load over-occurring 11-mers from 203 -ooc=11.ooc tells the program to load over-occurring 11-mers from
140 an external file. This will increase the speed 204 an external file. This will increase the speed
141 by a factor of 40 in many cases, but is not required. 205 by a factor of 40 in many cases, but is not required.
142 output.psl is the name of the output file. 206 output.psl is the name of the output file.
143 207
144 documentation: 208 documentation:
145 ++++++++++++++ 209 ++++++++++++++
146 210
147 See Blat documentation (http://genome.ucsc.edu/goldenPath/help/blatSpec.html) 211 See Blat documentation (http://genome.ucsc.edu/goldenPath/help/blatSpec.html)
148 212
149 Source code: 213 Source code:
150 ++++++++++++ 214 ++++++++++++
151 215
152 http://hgdownload.cse.ucsc.edu/admin/exe/ 216 http://hgdownload.cse.ucsc.edu/admin/exe/
153 217
154 ]]></help> 218 ]]></help>
155 <citations> 219 <citations>
156 <citation type="doi">10.1101/gr.229202</citation> 220 <citation type="doi">10.1101/gr.229202</citation>
157 </citations> 221 </citations>
158 </tool> 222 </tool>
159
160
161
162
163