|
0
|
1 <tool id="overlapselect_wrapper" name="overlap Select" version="1.0">
|
|
|
2 <requirements>
|
|
|
3 <requirement type='package' version="latest">overlapSelect</requirement>
|
|
|
4 </requirements>
|
|
|
5
|
|
|
6
|
|
|
7 <description>UCSC Overlap Select</description>
|
|
|
8 <command interpreter="bash">
|
|
|
9 overlapselect_wrapper.sh
|
|
|
10 #if $optionSpec.parType == "specify":
|
|
|
11 #if $str($optionSpec.strand) == "true":
|
|
|
12 -strand
|
|
|
13 #end if
|
|
|
14 #if $str($optionSpec.oppositeStrand) == "true":
|
|
|
15 -oppositeStrand
|
|
|
16 #end if
|
|
|
17 #if $str($optionSpec.nonOverlapping) == "true":
|
|
|
18 -nonOverlapping
|
|
|
19 #end if
|
|
|
20 #if $str($optionSpec.mergeOutput) == "true":
|
|
|
21 -mergeOutput
|
|
|
22 #end if
|
|
|
23 -overlapBases=$optionSpec.overlapBases
|
|
|
24 -overlapThreshold=$optionSpec.overlapThreshold
|
|
|
25 -overlapThresholdCeil=$optionSpec.overlapThresholdCeil
|
|
|
26 #if $str($optionSpec.selectFormat) == "bed":
|
|
|
27 -selectFmt=bed
|
|
|
28 #end if
|
|
|
29 #if $str($optionSpec.inFormat) == "bed":
|
|
|
30 -inFmt=bed
|
|
|
31 #end if
|
|
|
32 #if $str($optionSpec.selectFormat) == "tab":
|
|
|
33 -selectCoordCols=$optionSpec.selColStart
|
|
|
34 #end if
|
|
|
35 #if $str($optionSpec.inFormat) == "tab":
|
|
|
36 -inCoordCols=$optionSpec.inColStart
|
|
|
37 #end if
|
|
|
38 #else
|
|
|
39 -selectFmt=bed inFmt=bed
|
|
|
40 #end if
|
|
|
41 $selectFile
|
|
|
42 $inFile
|
|
|
43 $outFile
|
|
|
44 </command>
|
|
|
45 <inputs>
|
|
|
46 <param name="inFile" type="data" format="Tabular" metadata_name="dbkey" label="Select the query (input) file" />
|
|
|
47 <param name="selectFile" type="data" format="Tabular" metadata_name="dbkey" label="Select the reference (select) file" />
|
|
|
48 <conditional name="optionSpec">
|
|
|
49 <param name="parType" type="select" label="Do you want to select your target from the list or from your history?" help="">
|
|
|
50 <option value="default">Use Default Parameters</option>
|
|
|
51 <option value="specify">Specify Parameters</option>
|
|
|
52 </param>
|
|
|
53 <when value="default">
|
|
|
54 </when>
|
|
|
55 <when value="specify">
|
|
|
56 <param name="selectFormat" type="select" label="Query File Format -selectFmt">
|
|
|
57 <option value="bed">Bed</option>
|
|
|
58 <option value="tab">Tabular</option>
|
|
|
59 </param>
|
|
|
60 <param name="inFormat" type="select" label="Reference File Format -inFmt">
|
|
|
61 <option value="bed">Bed</option>
|
|
|
62 <option value="tab">Tabular</option>
|
|
|
63 </param>
|
|
|
64 <param name="selColStart" type="integer" value="0" label="Query File is unstranded with chr starting at this column (0 based - i.e. 0=col1), followed by start and end (-1 to disable)-selectCoordCols" />
|
|
|
65 <param name="inColStart" type="integer" value="0" label="Reference File is unstranded with chr starting at this column (0 based - i.e. 0=col1), followed by start and end (-1 to disable)-inCoordCols" />
|
|
|
66 <param name="strand" type="boolean" truevalue="true" falsevalue="false" checked="false" label="Only search on same strand -strand" />
|
|
|
67 <param name="oppositeStrand" type="boolean" truevalue="true" falsevalue="false" checked="false" label="Only search on opposite strand -oppositeStrand" />
|
|
|
68 <param name="nonOverlapping" type="boolean" truevalue="true" falsevalue="false" checked="false" label="Select non-overlapping instead of overlapping records" />
|
|
|
69 <param name="mergeOutput" type="boolean" truevalue="true" falsevalue="false" checked="false" label="Merge Output -mergeOutput" />
|
|
|
70 <param name="overlapBases" type="integer" value="-1" label="Number of overlap minimum (-overlapBases) (-1 to disable)" />
|
|
|
71 <param name="overlapThreshold" type="float" value="0.0" label="Minimum fraction overlap" help="minimum fraction of an inFile record that must be overlapped by a single select record to be considered
|
|
|
72 overlapping. Note that this is only coverage by a single select record, not total coverage" />
|
|
|
73 <param name="overlapThresholdCeil" type="float" value="1.1" label="Maximum fraction overlap" help="select only inFile records with less than this amount of overlap with a single record, provided they are selected by other criteria." />
|
|
|
74
|
|
|
75 </when> <!-- history -->
|
|
|
76 </conditional> <!-- optionSpec -->
|
|
|
77 </inputs>
|
|
|
78 <outputs>
|
|
|
79 <data format="interval" name="outFile" label="Overlapselect on ${selectFile.name} vs ${inFile.name} "/>
|
|
|
80 </outputs>
|
|
|
81 <tests>
|
|
|
82 <test>
|
|
|
83 <param name="inFile" value="NM_001206.gp"/>
|
|
|
84 <param name="selectFile" value="mrna.psl"/>
|
|
|
85 <output name="outFile" value="psl_over_NM_001206gp.psl"/>
|
|
|
86 </test>
|
|
|
87 </tests>
|
|
|
88
|
|
|
89 <help>
|
|
|
90 **Overlap Select Help**
|
|
|
91
|
|
|
92
|
|
|
93 **Options**
|
|
|
94
|
|
|
95 * selectCds - Use only CDS in the selectFile
|
|
|
96
|
|
|
97 * selectRange - Use entire range instead of blocks from records in the selectFile.
|
|
|
98
|
|
|
99 * inFmt=fmt - specify inFile format, same values as -selectFmt.
|
|
|
100
|
|
|
101 * inCoordCols=spec - inFile is tab-separate with coordinates specified by spec, in format described above.
|
|
|
102
|
|
|
103 * inCds - Use only CDS in the inFile
|
|
|
104
|
|
|
105 * inRange - Use entire range instead of blocks of records in the inFile.
|
|
|
106
|
|
|
107 * nonOverlapping - select non-overlapping instead of overlapping records
|
|
|
108
|
|
|
109 * strand - must be on the same strand to be considered overlapping
|
|
|
110
|
|
|
111 * oppositeStrand - must be on the opposite strand to be considered overlapping
|
|
|
112
|
|
|
113 * excludeSelf - don't compare records with the same coordinates and name. Warning: using only one of -inCds or -selectCds will result in different coordinates for the same record.
|
|
|
114
|
|
|
115 * idMatch - only select overlapping records if they have the same id
|
|
|
116
|
|
|
117 * aggregate - instead of computing overlap bases on individual select entries, compute it based on the total number of inFile bases overlap by selectFile records. -overlapSimilarity and -mergeOutput will not work with this option.
|
|
|
118
|
|
|
119 * overlapThreshold=0.0 - minimum fraction of an inFile record that must be overlapped by a single select record to be considered overlapping. Note that this is only coverage by a single select record, not total coverage.
|
|
|
120
|
|
|
121 * overlapThresholdCeil=1.1 - select only inFile records with less than this amount of overlap with a single record, provided they are selected by other criteria.
|
|
|
122
|
|
|
123 * overlapSimilarity=0.0 - minimum fraction of inFile and select records that: Note that this is only coverage by a single select record and this is; bidirectional inFile and selectFile must overlap by this amount. A value of 1.0 will select identical records (or CDS if both CDS options are specified. Not currently supported with *aggregate.*
|
|
|
124
|
|
|
125 * overlapSimilarityCeil=1.1 - select only inFile records with less than this amount of similarity with a single record. provided they are selected by other criteria.
|
|
|
126
|
|
|
127 * overlapBases=-1 - minimum number of bases of overlap, lt 0 disables.
|
|
|
128
|
|
|
129 * statsOutput - output overlap statistics instead of selected records. If no overlap criteria is specified, all overlapping entries are reported, Otherwise only the pairs passing the criteria are reported. This results in a tab-separated file with the columns: inId selectId inOverlap selectOverlap overBases: Where inOverlap is the fraction of the inFile record overlapped by the selectFile record and selectOverlap is the fraction of the select record overlap by inFile records. With -aggregate, output is: inId inOverlap inOverBases inBases
|
|
|
130
|
|
|
131 * statsOutputAll - like -statsOutput, however output all inFile records, including those that are not overlapped.
|
|
|
132
|
|
|
133 * statsOutputBoth - like -statsOutput, however output all selectFile and inFile records, including those that are not overlapped.
|
|
|
134
|
|
|
135 * mergeOutput - output file with be a merge of the input file with the selectFile records that selected it. The format is inRec selectRec. if multiple select records hit, inRec is repeated. This will increase
|
|
|
136 the memory required. Not supported with -nonOverlapping or -aggregate.
|
|
|
137
|
|
|
138 * idOutput - output a tab-separated file of pairs of inId selectId with -aggregate, only a single column of inId is written
|
|
|
139
|
|
|
140 * dropped=file - output rows that were dropped to this file.
|
|
|
141
|
|
|
142 * verbose=n - verbose gt 1 prints some details,
|
|
|
143
|
|
|
144
|
|
|
145 </help>
|
|
|
146 </tool>
|
|
|
147
|
