|
0
|
1 <tool name="Multiple wiggle files correlation in given regions" id="correlation_dataset_bed" version="0.1.0">
|
|
|
2 <description>Calculate the correlation coefficient on the genome scale using one bed file and multiple wiggle / bigwig files</description>
|
|
|
3 <macros>
|
|
|
4 <import>corr_macros.xml</import>
|
|
|
5 </macros>
|
|
|
6 <expand macro="requirements" />
|
|
|
7 <command>
|
|
|
8 #if $maxscore and $minscore and ( $minscore > $maxscore ):
|
|
|
9 echo "maxscore has to be greater than minscore!"
|
|
|
10 exit 1;
|
|
|
11 #end if
|
|
|
12 #if $wfile1.extension == "wig"
|
|
|
13 wig_correlation_in_bed_file.py -d $wfile1.metadata.dbkey -m $method
|
|
|
14 #elif $wfile1.extension == "bigwig"
|
|
|
15 bigwig_correlation_in_bed_file.py
|
|
|
16 #end if
|
|
|
17 #if $maxscore.value
|
|
|
18 --max-score $maxscore
|
|
|
19 #end if
|
|
|
20 #if $minscore.value
|
|
|
21 --min-score $minscore
|
|
|
22 #end if
|
|
|
23 $heatmap -z $size -f PDF -r tmp.r
|
|
|
24 -b $bfile
|
|
|
25 -w $wfile1 -l '$wlabel1' -w $wfile2 -l '$wlabel2'
|
|
|
26 #for $m in $more
|
|
|
27 -w $m.wig -l "$m.label"
|
|
|
28 #end for
|
|
|
29 &> $log &&
|
|
|
30 cp tmp.r $rscript &&
|
|
|
31 #if $imagetype.__str__ == "PNG":
|
|
|
32 convert tmp.r.pdf tmp.r.png &&
|
|
|
33 cp tmp.r.png $output
|
|
|
34 #else
|
|
|
35 cp tmp.r.pdf $output
|
|
|
36 #end if
|
|
|
37 </command>
|
|
|
38 <inputs>
|
|
|
39 <param name="bfile" type="data" format="bed" label="BED file(100,000 lines max)" >
|
|
|
40 <validator type="unspecified_build" />
|
|
|
41 </param>
|
|
|
42 <param format="wig,bigwig" name="wfile1" type="data" label="Wiggle / bigwig file 1">
|
|
|
43 <validator type="unspecified_build" />
|
|
|
44 </param>
|
|
|
45 <param name="wlabel1" type="text" label="Wiggle / bigwig label 1" optional="false">
|
|
|
46 <validator type="length" message="must be between 1 and 255 characters" min="1" max="255"/>
|
|
|
47 </param>
|
|
|
48 <param format="wig,bigwig" name="wfile2" type="data" label="Wiggle / bigwig file 2">
|
|
|
49 <validator type="unspecified_build" />
|
|
|
50 </param>
|
|
|
51 <param name="wlabel2" type="text" label="Wiggle / bigwig label 2" optional="false">
|
|
|
52 <validator type="length" message="must be between 1 and 255 characters" min="1" max="255"/>
|
|
|
53 </param>
|
|
|
54 <repeat name="more" title="wiggle / bigwig file">
|
|
|
55 <param format="wig,bigwig" name="wig" type="data" label="Select another wiggle / bigwig file"/>
|
|
|
56 <param name="label" type="text" label="Wiggle / bigwig label" optional="false">
|
|
|
57 <validator type="length" message="must be between 1 and 255 characters" min="1" max="255"/>
|
|
|
58 </param>
|
|
|
59 </repeat>
|
|
|
60 <param name="method" type="select" label="method:" help="method to process the paired two sets of data in the sampling step." >
|
|
|
61 <option value="mean" selected="true">mean</option>
|
|
|
62 <option value="median">median</option>
|
|
|
63 <option value="sample">sample</option>
|
|
|
64 </param>
|
|
|
65 <param name="maxscore" type="float" label="Max Score" help="Maximum score in the calculation (optional)" value="10000" optional="true"/>
|
|
|
66 <param name="minscore" type="float" label="Min Score" help="Minimum score in the calculation (optional)" value="-10000" optional="true"/>
|
|
|
67 <param name="heatmap" type="boolean" label="Draw Heatmap instead?" checked="no"
|
|
|
68 truevalue="-H" falsevalue=" " />
|
|
|
69 <param name="size" type="integer" label="image Size" value="10" help="Image size in inch. It must be ge 5 and le 20. Default: 10">
|
|
|
70 <validator type="in_range" max="20" min="5" message="image Size is out of range, image Size has to be between 5 to 20" />
|
|
|
71 </param>
|
|
|
72 <param type="select" name="imagetype" display="radio" label="Image Type">
|
|
|
73 <option value="PDF">PDF format</option>
|
|
|
74 <option value="PNG">PNG format</option>
|
|
|
75 </param>
|
|
|
76 </inputs>
|
|
|
77 <outputs>
|
|
|
78 <data format="png" name="output">
|
|
|
79 <change_format>
|
|
|
80 <when input="imagetype" value="PDF" format="pdf" />
|
|
|
81 </change_format>
|
|
|
82 </data>
|
|
|
83 <data format="txt" name="log" label="correlation job log" />
|
|
|
84 <data format="txt" name="rscript" label="correlation job rscript" />
|
|
|
85 </outputs>
|
|
|
86 <expand macro="stdio"/>
|
|
|
87 <tests>
|
|
|
88 <test>
|
|
|
89 <param name="bfile" value="peaks.bed" />
|
|
|
90 <param name="wfile1" value="control.wig" />
|
|
|
91 <param name="wlabel1" value="control" />
|
|
|
92 <param name="wfile2" value="treatment.wig" />
|
|
|
93 <param name="wlabel2" value="treatment" />
|
|
|
94 <param name="step" value="100" />
|
|
|
95 <param name="method" value="mean" />
|
|
|
96 <param name="maxscore" value="10000" />
|
|
|
97 <param name="minscore" value="-10000" />
|
|
|
98 <param name="size" value="10" />
|
|
|
99 <param name="imagetype" value="PDF" />
|
|
|
100 <output name="log">
|
|
|
101 <assert_contents>
|
|
|
102 <has_text_matching expression="Please check tmp.r" />
|
|
|
103 </assert_contents>
|
|
|
104 </output>
|
|
|
105 </test>
|
|
|
106 </tests>
|
|
|
107 <help>
|
|
|
108 This tool calculates the correlation coefficient values on the genome
|
|
|
109 scale using many scores datasets (many wiggle files) within the
|
|
|
110 regions given by a BED file. The tool is written by Tao Liu. It calls
|
|
|
111 R for plotting.
|
|
|
112
|
|
|
113 .. class:: infomark
|
|
|
114
|
|
|
115 **TIP:** This can be used to evaluate the correlation between
|
|
|
116 biological replicates at union peak regions.
|
|
|
117
|
|
|
118 .. class:: warningmark
|
|
|
119
|
|
|
120 **NEED IMPROVEMENT**
|
|
|
121
|
|
|
122 -----
|
|
|
123
|
|
|
124 **Parameters**
|
|
|
125
|
|
|
126 - **BED file** contains the regions to extract scores.
|
|
|
127 - **Wiggle file 1 and 2** are the first two wiggle files to be
|
|
|
128 included. These two are required.
|
|
|
129 - **Wiggle label 1 and 2** are the first two wiggle labels to be
|
|
|
130 plotted in the correlation plot.
|
|
|
131 - **wiggle files** click *Add new wiggle file* to add more wiggle
|
|
|
132 files and labels.
|
|
|
133 - **Genome/Assembly** Genome assembly to be used. The tool will
|
|
|
134 download the chromosome information from UCSC database.
|
|
|
135 - **Method** When scores are extracted for a region in BED file, a
|
|
|
136 method will be applied to calculate a value to represent this
|
|
|
137 region. Options are *median* to use the median value or *mean* to
|
|
|
138 use the average value.
|
|
|
139 - **Max Score** is the maximum score in the calculation. It's
|
|
|
140 optional.
|
|
|
141 - **Min Score** is the minimum score in the calculation. It's
|
|
|
142 optional.
|
|
|
143
|
|
|
144 -----
|
|
|
145
|
|
|
146 **Outputs**
|
|
|
147
|
|
|
148 - **PNG file** is the correlation plot
|
|
|
149 - **LOG file** for job log. If you see errors, please attached this in
|
|
|
150 the bug report
|
|
|
151
|
|
|
152 </help>
|
|
|
153
|
|
|
154 </tool>
|
