view data_manager/add_ctat_resource_lib.xml @ 27:c6a6d10bb2ad draft

Fix detect errors.

<tool id="ctat_genome_resource_libs_data_manager" 
    name="CTAT Genome Resource Libraries Data Manager" 
    version="2.0.0" tool_type="manage_data">
    <description>Retrieve, and/or specify the location of, a CTAT Genome Resource Library. 
    </description>
    <requirements>
        <requirement type="package" version="2.7">python</requirement>
        <requirement type="package" version="0.5.0">fusion-filter</requirement>
        <requirement type="package" version="2.0.1">ctat-mutations</requirement>
    </requirements>
    <command detect_errors="exit_code">
        <![CDATA[
          python $__tool_directory__/add_ctat_resource_lib.py 
            --output_filename="${out_file}" 
            --display_name="${display_name}"            
            #if str( $genome_resource_library.build_type ) == "download_and_build":
              --download_url="${genome_resource_library.download_url}" 
              --download_location="${genome_resource_library.download_destination}"
              #if str( $genome_resource_library.force_new_download ) == "true":
                --new_archive_download
              #end if
              #if str( $genome_resource_library.keep_archive ) == "true":
                --keep_archive
              #end if
              #if str( $genome_resource_library.rebuild ) == "true":
                --new_library_build
              #end if
              #if str( $genome_resource_library.specify_build_location.build_location ) == "different_location":
                  --build_location="${genome_resource_library.specify_build_location.different_build_location}"
              #end if
            #elif str( $genome_resource_library.build_type ) == "build_from_source":
              --source_location "${genome_resource_library.source_location}"
              #if str( $genome_resource_library.rebuild ) == "true":
                --new_library_build
              #end if
              #if str( $genome_resource_library.specify_build_location.build_location ) == "true":
                  --build_location="${genome_resource_library.specify_build_location.different_build_location}"
              #end if
            #elif str( $genome_resource_library.build_type ) == "specify_built_location":
              --build_location="${genome_resource_library.built_library_location}"
            #end if
            
            #if str( $gmap_options.gmap_build ) == "true":
              --gmap_build 
              #if str( $gmap_options.force_gmap_build ) == "true":
                  --force_gmap_build
              #end if
            #end if
            
            #if str( $mutation_lib_options.mutation_build ) == "true":
              --download_mutation_resources_url="${mutation_lib_options.source_url}" 
              --cosmic_resources_location="${mutation_lib_options.cosmic_files_location}"
              #if str( $mutation_lib_options.force_download ) == "true":
                  --new_mutation_download
              #end if
              #if str( $mutation_lib_options.redo_integration ) == "true":
                  --new_mutation_integration
              #end if
            #end if
        ]]>
    </command>
    <inputs>
        <!-- The following are left in here, just as examples of various ways of doing options.
            <param name="force_download" type="boolean" checked="false"
                truevalue="- -force_download" falsevalue="" label="Force New Download? (yes/no)" />
            <param name="download" type="select" label="Need to Download?">
                <option value="single" selected="true">Single Dataset</option>
                <option value="paired_collection">Paired Collection</option>
            <when value="paired_collection">
                 <param name="fastq_input" format="fastqsanger" type="data_collection" collection_type="paired" label="Select dataset pair" help="Specify paired dataset collection containing paired reads"/>
            </when>
        -->
        <conditional name="genome_resource_library">
            <param name="build_type" type="select" label="Download CTAT Genome Resource Library?">
                <option value="download_and_build" selected="true">Download from CTAT and build if needed</option>
                <option value="build_from_source">Build library from local source data</option>
                <option value="specify_built_location">Specify location of built library</option>
            </param>
            <when value="download_and_build">
                <!-- The use of a code block to get dynamic options is now deprecated and discouraged.
                     I am still using it here. The only other way I can think of to do this is to
                     create another data_manager that gets the list of files and puts them into a
                     data_table, that is then used to get the filenames. That would require the admin
                     to first run the data_manager that builds the filename data_table before running
                     this data_manager.
                This is the dynamic way to get the options filled.
                <param name="filename" type="select" label="Select File" display="radio" 
                    dynamic_options="get_ctat_genome_filenames()" 
                    help="Select a CTAT Genome Resource Library to Download." />
                Here is the static method for what is online in April 2017:
                <param name="filename" type="select" label="Choose which library to download.">
                    <option value="https://data.broadinstitute.org/Trinity/CTAT_RESOURCE_LIB/GRCh37_v19_CTAT_lib_Feb092018.plug-n-play.tar.gz">
                        GRCh37_v19_CTAT_lib_Feb092018.plug-n-play.tar.gz
                    </option>
                    <option value="https://data.broadinstitute.org/Trinity/CTAT_RESOURCE_LIB/                        GRCh37_v19_CTAT_lib_Feb092018.source_data.tar.gz">
                        GRCh37_v19_CTAT_lib_Feb092018.source_data.tar.gz
                    </option>                        
                    <option value="https://data.broadinstitute.org/Trinity/CTAT_RESOURCE_LIB/                        GRCh38_v27_CTAT_lib_Feb092018.plug-n-play.tar.gz">
                        GRCh38_v27_CTAT_lib_Feb092018.plug-n-play.tar.gz
                    </option>
                    <option value="https://data.broadinstitute.org/Trinity/CTAT_RESOURCE_LIB/                        GRCh38_v27_CTAT_lib_Feb092018.source_data.tar.gz">
                        GRCh38_v27_CTAT_lib_Feb092018.source_data.tar.gz
                    </option>
                    <option value="https://data.broadinstitute.org/Trinity/CTAT_RESOURCE_LIB/                        Mouse_M16_CTAT_lib_Feb202018.plug-n-play.tar.gz">
                        Mouse_M16_CTAT_lib_Feb202018.plug-n-play.tar.gz
                    </option>
                    <option value="https://data.broadinstitute.org/Trinity/CTAT_RESOURCE_LIB/                        Mouse_M16_CTAT_lib_Feb202018.source_data.tar.gz">
                        Mouse_M16_CTAT_lib_Feb202018.source_data.tar.gz
                    </option>
                -->
                <param name="download_url" type="select" label="Select a File"
                    dynamic_options="get_ctat_genome_urls()" 
                    help="Select a CTAT Genome Resource Library to Download.">
                </param>
                <param name="download_destination" type="text" label="Download Destination (full path)" />
                <param name="force_new_download" type="boolean" checked="false" label="Force New Download?" />
                <param name="keep_archive" type="boolean" checked="true" label="Keep downloaded archive rather than deleting after build?" />
                <param name="rebuild" type="boolean" checked="false" label="Force new build of Library?" />
                <conditional name="specify_build_location">
                    <param name="build_location" type="select" label="Location of Built Library">
                        <option value="same_as_download" selected="true">Build in Download location</option>
                        <option value="different_location">Build in a different location</option>
                    </param>
                    <when value="same_as_download">
                        <!-- do nothing here -->
                    </when>
                    <when value="different_location">
                        <param name="different_build_location" type="text" label="Build Location (full path)" />
                    </when>
                </conditional>
            </when>
            <when value="build_from_source">
                <param name="source_location" type="text" label="Location of Source Files (full path)" />
                <param name="rebuild" type="boolean" checked="false" label="Force new build of Library?" />
                <conditional name="specify_build_location">
                    <param name="build_location" type="boolean" checked="false" label="Is the build location different than the source location?" />
                    <when value="true">
                        <param name="different_build_location" type="text" label="Build Location (full path)" />
                    </when>
                </conditional>
            </when>
            <when value="specify_built_location">
                <param name="built_library_location" type="text" label="Location of the Built Library (full path)" />
            </when>
        </conditional>
        <param name="display_name" type="text" label="Reference Genome Display Name" />
        <conditional name="gmap_options">
            <param name="gmap_build" type="boolean" checked="true" label="Do a gmap_build on the Library?" />
            <when value="true">
                <param name="force_gmap_build" type="boolean" checked="false" label="Force a new build even if previously done?" />
            </when>
        </conditional>
        <conditional name="mutation_lib_options">
            <param name="mutation_build" type="boolean" checked="false" label="Download mutation library?">
            </param>
            <when value="true">
                <param name="source_url" type="select" label="Select a File"
                    dynamic_options="get_mutation_resource_urls()" 
                    help="Select CTAT Mutation Library File to Download.\nMake sure it is the right one for your CTAT Genome Resource Library!">
                </param>
                <param name="force_download" type="boolean" checked="false" label="Force New Download?" />
                <param name="redo_integration" type="boolean" checked="false" label="Redo Library Integration?" />
                <param name="cosmic_files_location" type="text" label="Location of the COSMIC files (See Tool Notes)." />
            </when>
        </conditional>
        -->
    </inputs>
    <outputs>
        <data name="out_file" format="data_manager_json" />
    </outputs>
    <help>
        Retrieve, and/or specify the location of, a CTAT Genome Resource Library.
        When download is true, the files at https://data.broadinstitute.org/Trinity/CTAT_RESOURCE_LIB/
        are used as selectors for the user to choose among.
        Specify the Full Path of the location where the CTAT Genome Resource Library should be placed.
        You will need approximately 62GB of space for this library.
        The installation of this tool takes some time, due to building a conda environment for the dependencies.
        The download extracts the files during the download. The "source_data" files download faster, but then must be built.
        Building the library from the "source_data" files can take many hours, depending on the resources of your machine.
        The "plug-n-play" can take considerable time to download, depending on your internet connection. Even with high speed,
        it is about 25GB that is transfered, so plan accordingly.
        If you have a good speed internet connection, downloading the plug-n-play will usually be faster than building.
        If a download or the build is interupted, re-running the job should pick up where it left off.
        Neither the "source_data" nor the "plug-n-play" versions have had their gmap index built. If you are not going to be
        using gmap_fusion, then you can uncheck the gmap_build check box and save the space and time building the index consumes.
        Neither the "source_data" nor the "plug-n-play" versions have mutation resources included. Those must be downloaded
        separately. By default the Mutation Resources are not integrated into the Library. If you are going to be using the 
        ctat_mutations tool, check the Download Mutation Library check box. 
        In order to integrate the Mutation Resources into a CTAT Genome Resource Library, you must have previously downloaded
        COSMIC resources (See Step 2 from 
        <a href="https://github.com/NCIP/ctat-mutations/tree/master/mutation_lib_prep">Mutation Lib Prep Information</a>.)
        You can place them directly into the Genome Resource Library location, or if the Library is 
        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 
        integrated into the Library. The Mouse genome is not currently supported in ctat_mutations.
        If you already have a CTAT Genome Resource library installed on your system, 
        specify the full path of the location where it exists and leave the download box unchecked.
        The Reference Genome name may be left empty if downloading. 
        The filename will then be used as the selector text of the entry in the data table.
        For more information on CTAT Genome Resource Libraries, 
        see <a href="https://github.com/FusionFilter/FusionFilter/wiki">FusionFilter</a>
    </help>
    <code file="add_ctat_resource_lib.py" />
</tool>