Mercurial > repos > iuc > umi_tools_extract

comparison umi-tools_extract.xml @ 5:f77bc14eba31 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload commit 57e3e460a740aa7aad217c8365527c49e88c9a30
author iuc
date Tue, 05 Jun 2018 19:44:38 -0400
parents e73a22ff585c
children 3cfd8e1073d7
comparison
equal deleted inserted replaced
4:e73a22ff585c 5:f77bc14eba31
6 <expand macro="requirements" /> 6 <expand macro="requirements" />
7 <command detect_errors="exit_code"><![CDATA[ 7 <command detect_errors="exit_code"><![CDATA[
8 @COMMAND_LINK@ 8 @COMMAND_LINK@
9 9
10 umi_tools extract 10 umi_tools extract
11 --extract-method='$extract_method.value'
11 --bc-pattern='$bc_pattern' 12 --bc-pattern='$bc_pattern'
13
12 #if $input_type.type == 'single': 14 #if $input_type.type == 'single':
13 #if $gz: 15 #if $gz:
14 --stdin=input_single.gz 16 --stdin=input_single.gz
15 --stdout out.gz 17 --stdout out.gz
16 #else 18 #else
17 --stdin='$input_type.input_single' 19 --stdin=input_single.txt
18 --stdout '$out' 20 --stdout '$out'
19 #end if 21 #end if
20 #else: 22 #else:
21 #if $gz: 23 #if $gz:
22 --stdin=input_read1.gz 24 --stdin=input_read1.gz
23 --read2-in=input_read2.gz 25 --read2-in=input_read2.gz
24 --stdout out1.gz 26 --stdout out1.gz
25 --read2-out=out2.gz 27 --read2-out=out2.gz
26 #else: 28 #else:
27 --stdin='$input_type.input_read1' 29 --stdin=input_read1.txt
28 --read2-in='$input_type.input_read2' 30 --read2-in=input_read2.txt
29 --stdout '$out1' 31 --stdout '$out1'
30 --read2-out='$out2' 32 --read2-out='$out2'
31 #end if 33 #end if
32 #if $input_type.barcode.barcode_select == "both_reads": 34 #if $input_type.barcode.barcode_select == "both_reads":
33 --split-barcode 35 --split-barcode
34 --bc-pattern2='$input_type.barcode.bc_pattern2' 36 --bc-pattern2='$input_type.barcode.bc_pattern2'
35 #end if 37 #end if
36 #end if 38 #end if
39
40 #if $barcodes.use_barcodes.value == 'yes':
41 --filter-cell-barcode
42 --whitelist='$barcodes.filter_barcode_file'
43 '$barcodes.filter_correct.value'
44 #end if
45
37 #if not $prime3: 46 #if not $prime3:
38 --3prime 47 --3prime
39 #end if 48 #end if
40 #if $quality.quality_selector =='true': 49 #if $quality.quality_selector =='true':
41 --quality-filter-threshold '$quality.quality_filter_threshold' 50 --quality-filter-threshold '$quality.quality_filter_threshold'
55 #end if 64 #end if
56 #end if 65 #end if
57 ]]></command> 66 ]]></command>
58 <inputs> 67 <inputs>
59 <expand macro="input_types" /> 68 <expand macro="input_types" />
69
70 <conditional name="barcodes" >
71 <param name="use_barcodes" argument="--filter-cell-barcode" type="select" label="Use Known Barcodes?" >
72 <option value="yes">Yes</option>
73 <option value="no" selected="true" >No</option>
74 </param>
75 <when value="no" />
76 <when value="yes" >
77 <param name="filter_barcode_file" type="data" format="tsv" label="Barcode File" />
78 <param name="filter_correct" argument="--error-correct-cell" type="boolean" truevalue="--error-correct-cell" falsevalue="" checked="false" label="Apply correction to cell barcodes?" help="This only applies if your barcode file has two columns output from the umi_tools whitelist command." />
79 </when>
80 </conditional>
81
82 <param name="extract_method" type="select" label="Method to extract barcodes" >
83 <option value="regex">Regular Expressions</option>
84 <option value="string" selected="true">String</option>
85 </param>
86
60 <param name="bc_pattern" argument="--bc-pattern" type="text" label="Barcode pattern for first read" 87 <param name="bc_pattern" argument="--bc-pattern" type="text" label="Barcode pattern for first read"
61 help="Use this option to specify the format of the UMI/barcode. Use Ns to 88 help="Use this option to specify the format of the UMI/barcode. Use Ns to
62 represent the random positions and Xs to indicate the bc positions. 89 represent the random positions and Xs to indicate the bc positions.
63 Bases with Ns will be extracted and added to the read name. Remaining 90 Bases with Ns will be extracted and added to the read name. Remaining
64 bases, marked with an X will be reattached to the read."> 91 bases, marked with an X will be reattached to the read.">
92 <sanitizer invalid_char="">
93 <valid initial="string.letters,string.digits">
94 <add value="!="/>
95 <add value="-"/>
96 <add value="_"/>
97 <add value="."/>
98 <add value="?"/>
99 <add value="&lt;"/><!-- left triangle bracket -->
100 <add value="&gt;"/><!-- right triangle bracket -->
101 <add value="&#91;"/> <!-- left square bracket -->
102 <add value="&#93;"/> <!-- right square bracket -->
103 <add value="&#94;"/> <!-- caret -->
104 <add value="&#123;"/> <!-- left curly -->
105 <add value="&#125;"/> <!-- right curly -->
106 <add value="&#40;"/> <!-- left parenthesis -->
107 <add value="&#41;"/> <!-- right parenthesis -->
108 </valid>
109 </sanitizer>
65 </param> 110 </param>
111
66 <param name="prime3" argument="--3prime" type="boolean" label="Is the barcode at the 5' end?" 112 <param name="prime3" argument="--3prime" type="boolean" label="Is the barcode at the 5' end?"
67 truevalue="1" falsevalue="0" checked="true" 113 truevalue="1" falsevalue="0" checked="true"
68 help="By default the barcode is assumed to be on the 5' end of the read, but 114 help="By default the barcode is assumed to be on the 5' end of the read, but
69 use this option to sepecify that it is on the 3' end instead." /> 115 use this option to sepecify that it is on the 3' end instead." />
70 <param name="print_log" argument="-L" type="boolean" label="Output log?" 116 <param name="print_log" argument="-L" type="boolean" label="Output log?"
124 <output name="out1" file="out_R1.fastq.gz" decompress="true" lines_diff="2" /> 170 <output name="out1" file="out_R1.fastq.gz" decompress="true" lines_diff="2" />
125 <output name="out2" file="out_R2.fastq.gz" decompress="true" lines_diff="2" /> 171 <output name="out2" file="out_R2.fastq.gz" decompress="true" lines_diff="2" />
126 <output name="out_log" file="out_paired.log" lines_diff="16"/> 172 <output name="out_log" file="out_paired.log" lines_diff="16"/>
127 </test> 173 </test>
128 <test> 174 <test>
129 <param name="type" value="paired_collection" /> 175 <param name="type" value="paired_collection" /> <!-- same as before, but uncompressed -->
130 <param name="input_readpair" > 176 <param name="input_readpair" >
131 <collection type="paired"> 177 <collection type="paired">
132 <element name="forward" ftype="fastq.gz" value="t_R1.fastq.gz" /> 178 <element name="forward" ftype="fastq" value="t_R1.fastq" />
133 <element name="reverse" ftype="fastq.gz" value="t_R2.fastq.gz" /> 179 <element name="reverse" ftype="fastq" value="t_R2.fastq" />
134 </collection> 180 </collection>
135 </param> 181 </param>
136 <param name="bc_pattern" value="NNNXXX" /> 182 <param name="bc_pattern" value="NNNXXX" />
137 <output name="out1" file="out_R1.fastq.gz" decompress="true" lines_diff="2" /> 183 <output name="out1" file="out_R1.fastq.gz" decompress="true" lines_diff="2" />
138 <output name="out2" file="out_R2.fastq.gz" decompress="true" lines_diff="2" /> 184 <output name="out2" file="out_R2.fastq.gz" decompress="true" lines_diff="2" />
139 <output name="out_log" file="out_paired.log" lines_diff="16"/> 185 <output name="out_log" file="out_paired.log" lines_diff="25" />
186 </test>
187 <test>
188 <param name="type" value="paired" />
189 <param name="input_read1" value="scrb_seq_fastq.1.gz" ftype="fastq.gz" />
190 <param name="input_read2" value="scrb_seq_fastq.2.gz" ftype="fastq.gz" />
191 <param name="extract_method" value="string" />
192 <param name="bc_pattern" value="CCCCCCNNNNNNNNNN" />
193 <param name="use_barcodes" value="yes" />
194 <param name="filter_barcode_file" value="scrb_seq_barcodes" />
195 <output name="out2" file="scrb_extract.fastq.gz" decompress="true" />
196 </test>
197 <test><!-- same as above but with regex barcode-->
198 <param name="type" value="paired" />
199 <param name="input_read1" value="scrb_seq_fastq.1.gz" ftype="fastq.gz" />
200 <param name="input_read2" value="scrb_seq_fastq.2.gz" ftype="fastq.gz" />
201 <param name="extract_method" value="regex" />
202 <param name="bc_pattern" value="^(?P&lt;cell_1&gt;.{6})(?P&lt;umi_1&gt;.{10})" />
203 <param name="use_barcodes" value="yes" />
204 <param name="filter_barcode_file" value="scrb_seq_barcodes" />
205 <output name="out2" file="scrb_extract.fastq.gz" decompress="true" />
140 </test> 206 </test>
141 </tests> 207 </tests>
142 <help><![CDATA[ 208 <help><![CDATA[
143 209
144 210