44
|
1 <tool id="ctat_genome_resource_libs_data_manager"
|
|
2 name="CTAT Genome Resource Libraries Data Manager"
|
|
3 version="2.0.0" tool_type="manage_data">
|
49
|
4 <!-- This Data Manager tool was written by Cicada Dennis of Indiana University for the Broad Institute.
|
|
5 -->
|
44
|
6 <description>Retrieve, and/or specify the location of, a CTAT Genome Resource Library.
|
|
7 </description>
|
|
8 <requirements>
|
|
9 <requirement type="package" version="2.7">python</requirement>
|
|
10 <requirement type="package" version="0.5.0">fusion-filter</requirement>
|
|
11 <requirement type="package" version="2.0.1">ctat-mutations</requirement>
|
|
12 </requirements>
|
|
13 <command detect_errors="exit_code">
|
|
14 <![CDATA[
|
|
15 python $__tool_directory__/add_ctat_resource_lib.py
|
|
16 --output_filename="${out_file}"
|
52
|
17 --display_name="${display_name}"
|
|
18 --cravat_tissues_filepath="${__tool_directory__}/../tool-data/ctat_cravat_tissues.loc.sample"
|
44
|
19 #if str( $genome_resource_library.build_type ) == "download_and_build":
|
|
20 --download_url="${genome_resource_library.download_url}"
|
|
21 --download_location="${genome_resource_library.download_destination}"
|
|
22 #if str( $genome_resource_library.force_new_download ) == "true":
|
|
23 --new_archive_download
|
|
24 #end if
|
|
25 #if str( $genome_resource_library.keep_archive ) == "true":
|
|
26 --keep_archive
|
|
27 #end if
|
|
28 #if str( $genome_resource_library.rebuild ) == "true":
|
|
29 --new_library_build
|
|
30 #end if
|
|
31 #if str( $genome_resource_library.specify_build_location.build_location ) == "true":
|
|
32 --build_location="${genome_resource_library.specify_build_location.different_build_location}"
|
|
33 #end if
|
|
34 #elif str( $genome_resource_library.build_type ) == "build_from_source":
|
|
35 --source_location "${genome_resource_library.source_location}"
|
|
36 #if str( $genome_resource_library.rebuild ) == "true":
|
|
37 --new_library_build
|
|
38 #end if
|
|
39 #if str( $genome_resource_library.specify_build_location.build_location ) == "true":
|
|
40 --build_location="${genome_resource_library.specify_build_location.different_build_location}"
|
|
41 #end if
|
|
42 #elif str( $genome_resource_library.build_type ) == "specify_built_location":
|
|
43 --build_location="${genome_resource_library.built_library_location}"
|
|
44 #end if
|
|
45
|
|
46 #if str( $gmap_options.gmap_build ) == "true":
|
|
47 --gmap_build
|
|
48 #if str( $gmap_options.force_gmap_build ) == "true":
|
|
49 --force_gmap_build
|
|
50 #end if
|
|
51 #end if
|
|
52
|
|
53 #if str( $mutation_lib_options.mutation_build ) == "true":
|
|
54 --download_mutation_resources_url="${mutation_lib_options.source_url}"
|
|
55 --cosmic_resources_location="${mutation_lib_options.cosmic_files_location}"
|
|
56 #if str( $mutation_lib_options.force_download ) == "true":
|
|
57 --new_mutation_download
|
|
58 #end if
|
|
59 #if str( $mutation_lib_options.redo_integration ) == "true":
|
|
60 --new_mutation_integration
|
|
61 #end if
|
|
62 #end if
|
|
63 ]]>
|
|
64 </command>
|
|
65 <inputs>
|
|
66 <conditional name="genome_resource_library">
|
|
67 <param name="build_type" type="select" label="Download CTAT Genome Resource Library?">
|
|
68 <option value="download_and_build" selected="true">Download from CTAT and build if needed</option>
|
|
69 <option value="build_from_source">Build library from local source data</option>
|
|
70 <option value="specify_built_location">Specify location of built library</option>
|
|
71 </param>
|
|
72 <when value="download_and_build">
|
|
73 <!-- The use of a code block to get dynamic options is now deprecated and discouraged.
|
|
74 I am still using it here. The only other way I can think of to do this is to
|
|
75 create another data_manager that gets the list of files and puts them into a
|
|
76 data_table, that is then used to get the filenames. That would require the admin
|
|
77 to first run the data_manager that builds the filename data_table before running
|
|
78 this data_manager.
|
|
79 This is the dynamic way to get the options filled.
|
|
80 <param name="filename" type="select" label="Select File" display="radio"
|
49
|
81 dynamic_options="get_ctat_genome_urls()"
|
44
|
82 help="Select a CTAT Genome Resource Library to Download." />
|
|
83 Here is the static method for what is online in April 2017:
|
|
84 <param name="filename" type="select" label="Choose which library to download.">
|
|
85 <option value="https://data.broadinstitute.org/Trinity/CTAT_RESOURCE_LIB/GRCh37_v19_CTAT_lib_Feb092018.plug-n-play.tar.gz">
|
|
86 GRCh37_v19_CTAT_lib_Feb092018.plug-n-play.tar.gz
|
|
87 </option>
|
|
88 <option value="https://data.broadinstitute.org/Trinity/CTAT_RESOURCE_LIB/ GRCh37_v19_CTAT_lib_Feb092018.source_data.tar.gz">
|
|
89 GRCh37_v19_CTAT_lib_Feb092018.source_data.tar.gz
|
|
90 </option>
|
|
91 <option value="https://data.broadinstitute.org/Trinity/CTAT_RESOURCE_LIB/ GRCh38_v27_CTAT_lib_Feb092018.plug-n-play.tar.gz">
|
|
92 GRCh38_v27_CTAT_lib_Feb092018.plug-n-play.tar.gz
|
|
93 </option>
|
|
94 <option value="https://data.broadinstitute.org/Trinity/CTAT_RESOURCE_LIB/ GRCh38_v27_CTAT_lib_Feb092018.source_data.tar.gz">
|
|
95 GRCh38_v27_CTAT_lib_Feb092018.source_data.tar.gz
|
|
96 </option>
|
|
97 <option value="https://data.broadinstitute.org/Trinity/CTAT_RESOURCE_LIB/ Mouse_M16_CTAT_lib_Feb202018.plug-n-play.tar.gz">
|
|
98 Mouse_M16_CTAT_lib_Feb202018.plug-n-play.tar.gz
|
|
99 </option>
|
|
100 <option value="https://data.broadinstitute.org/Trinity/CTAT_RESOURCE_LIB/ Mouse_M16_CTAT_lib_Feb202018.source_data.tar.gz">
|
|
101 Mouse_M16_CTAT_lib_Feb202018.source_data.tar.gz
|
|
102 </option>
|
|
103 -->
|
|
104 <param name="download_url" type="select" label="Select a File"
|
|
105 dynamic_options="get_ctat_genome_urls()"
|
|
106 help="Select a CTAT Genome Resource Library to Download.">
|
|
107 </param>
|
|
108 <param name="download_destination" type="text" label="Download Destination (full path)" />
|
|
109 <param name="force_new_download" type="boolean" checked="false" label="Force New Download?" />
|
|
110 <param name="keep_archive" type="boolean" checked="true" label="Keep downloaded archive rather than deleting after build?" />
|
|
111 <param name="rebuild" type="boolean" checked="false" label="Force new build of Library?" />
|
|
112 <conditional name="specify_build_location">
|
|
113 <param name="build_location" type="boolean" checked="false" label="Is the build location different than the source location?" />
|
|
114 <when value="true">
|
|
115 <param name="different_build_location" type="text" label="Build Location (full path)" />
|
|
116 </when>
|
|
117 </conditional>
|
|
118 </when>
|
|
119 <when value="build_from_source">
|
|
120 <param name="source_location" type="text" label="Location of Source Files (full path)" />
|
|
121 <param name="rebuild" type="boolean" checked="false" label="Force new build of Library?" />
|
|
122 <conditional name="specify_build_location">
|
|
123 <param name="build_location" type="boolean" checked="false" label="Is the build location different than the source location?" />
|
|
124 <when value="true">
|
|
125 <param name="different_build_location" type="text" label="Build Location (full path)" />
|
|
126 </when>
|
|
127 </conditional>
|
|
128 </when>
|
|
129 <when value="specify_built_location">
|
|
130 <param name="built_library_location" type="text" label="Location of the Built Library (full path)" />
|
|
131 </when>
|
|
132 </conditional>
|
|
133 <param name="display_name" type="text" label="Reference Genome Display Name" />
|
|
134 <conditional name="gmap_options">
|
|
135 <param name="gmap_build" type="boolean" checked="true" label="Do a gmap_build on the Library?" />
|
|
136 <when value="true">
|
|
137 <param name="force_gmap_build" type="boolean" checked="false" label="Force a new build even if previously done?" />
|
|
138 </when>
|
|
139 </conditional>
|
|
140 <conditional name="mutation_lib_options">
|
|
141 <param name="mutation_build" type="boolean" checked="false" label="Download and integrate mutation library?">
|
|
142 </param>
|
|
143 <when value="true">
|
|
144 <param name="source_url" type="select" label="Select a File"
|
|
145 dynamic_options="get_mutation_resource_urls()"
|
|
146 help="Select CTAT Mutation Library File to Download.\nMake sure it is the right one for your CTAT Genome Resource Library!">
|
|
147 </param>
|
|
148 <param name="force_download" type="boolean" checked="false" label="Force New Download?" />
|
|
149 <param name="redo_integration" type="boolean" checked="false" label="Redo Library Integration?" />
|
|
150 <param name="cosmic_files_location" type="text" label="Location of the COSMIC files (See Tool Notes)." />
|
|
151 </when>
|
|
152 </conditional>
|
|
153 </inputs>
|
|
154 <outputs>
|
|
155 <data name="out_file" format="data_manager_json" />
|
|
156 </outputs>
|
|
157 <help>
|
|
158 Retrieve, and/or specify the location of, a CTAT Genome Resource Library.
|
|
159
|
|
160 There are three options: 1) Download from CTAT and build if needed; 2) Build library from local source data; 3) Specify location of built library.
|
|
161
|
|
162 1) Download from CTAT and build if needed:
|
|
163 This option uses the files at https://data.broadinstitute.org/Trinity/CTAT_RESOURCE_LIB/
|
|
164 as selectors for the user to choose among.
|
|
165 Specify the Full Path of the location where the CTAT Genome Resource Library should be placed.
|
|
166 The other options should be self-explanatory.
|
|
167 2) Build library from local source data:
|
|
168 This assumes that you have already downloaded the source data for a build and you want to build it.
|
|
169 3) Specify location of built library:
|
|
170 This assumes you already have a build library and need tell galaxy where it is.
|
|
171
|
|
172 All options allow the user to do a gmap_build on the library
|
|
173 and also to integrate ctat-mutation resources into the library.
|
|
174
|
49
|
175 You will need approximately 62GB of space for a human genome resource library, once it is built,
|
44
|
176 but if downloading and building, to be safe provide at least 75GB.
|
|
177
|
|
178 The installation of this tool takes some time, due to building a conda environment for the dependencies.
|
|
179 The "source_data" files download faster, but then must be built.
|
|
180 Building the library from the "source_data" files can take many hours, depending on the resources of your machine.
|
|
181 The "plug-n-play" can take considerable time to download, depending on your internet connection. Even with high speed,
|
|
182 it is about 25GB that is transfered, so plan accordingly.
|
|
183 If you have a good speed internet connection, downloading the plug-n-play will usually be faster than building.
|
|
184
|
|
185 **If a download or a build is interrupted, re-running the job should pick up where it left off.**
|
|
186
|
|
187 Neither the "source_data" nor the "plug-n-play" versions have had their gmap index built. If you are not going to be
|
|
188 using gmap_fusion, then you can uncheck the gmap_build check box and save the space and time building the index consumes.
|
|
189
|
49
|
190 Neither the "source_data" nor the "plug-n-play" versions have Mutation Resources included.
|
|
191 Those must be downloaded separately and integrated into the Library. If you are going to be using the
|
52
|
192 ctat_mutations tool, check the Download Mutation Library check box. Whether or not you check this box, the
|
|
193 ctat_cravat_tissues table needed by the ctat_mutations tool will be created.
|
44
|
194
|
|
195 In order to integrate the Mutation Resources into a CTAT Genome Resource Library, you must have previously downloaded
|
|
196 COSMIC resources (See Step 2 from https://github.com/NCIP/ctat-mutations/tree/master/mutation_lib_prep )
|
|
197 You can place them directly into the Genome Resource Library location, or if the Library is
|
|
198 not built yet, or you do not know the full path to it, specify the directory where the COSMIC files are, so they can be
|
|
199 integrated into the Library.
|
|
200
|
|
201 The Mouse genome is not currently supported by ctat_mutations.
|
|
202
|
|
203 If the Reference Genome Display Name is left empty a name will be created,
|
|
204 but any text that will best guide the user can be entered here.
|
|
205 It will be the text that is used for selecting the library in pull down lists
|
49
|
206 requiring a Genome Reference Library resource (These are stored in the ctat_genome_resource_libs table).
|
44
|
207
|
|
208 For more information on CTAT Genome Resource Libraries,
|
|
209 see https://github.com/FusionFilter/FusionFilter/wiki
|
|
210 </help>
|
|
211 <code file="add_ctat_resource_lib.py" />
|
|
212 </tool>
|