view data_manager/add_ctat_resource_lib.xml @ 8:b2e6ed40840a draft

Uploaded

<tool id="ctat_genome_resource_libs_data_manager" 
    name="CTAT Genome Resource Libraries Data Manager" 
    version="1.0.0" 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>
    </requirements>
    <command detect_errors="default">
        <![CDATA[
        python $__tool_directory__/add_ctat_resource_lib.py 
            --display_name "${display_name}" 
            --destination_path "${destination}" 
            --output_filename "${out_file}" 
            #if str( $genome_resource_library.download ) == "true":
                --source_url "${genome_resource_library.source_url}" 
                #if str( $genome_resource_library.force_download ) == "true":
                    --force_download
                #end if
            #end if
            #if str( $rebuild ) == "true":
                --build 
            #end if
            #if str( $gmap_build ) == "true":
                --gmap_build 
            #end if
            #if str( $mutation_indexes.download ) == "true":
                --download_mutation_indexes "${mutation_indexes.source_url}" 
                #if str( $mutation_indexes.force_download ) == "true":
                    --force_mutation_indexes_download
                #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="download" type="boolean" checked="false" label="Download CTAT Genome Resource Library?">
            </param>
            <when value="true">
                <!-- 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="source_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="force_download" type="boolean" checked="false" label="Force New Download?" />
            </when>
        </conditional>

        <param name="display_name" type="text" label="Reference Genome Display Name" />
        <param name="destination" type="text" label="Local Destination (full path)" />
        <param name="rebuild" type="boolean" checked="false" label="Force rebuild of Library?" />
        <param name="gmap_build" type="boolean" checked="true" label="Do a gmap_build on the Library?" />
        <!-- <param name="mutation_indexes" type="boolean" checked="true" label="Download mutation indexes into the Library?" />
        -->
        <conditional name="mutation_indexes">
            <param name="download" type="boolean" checked="true" label="Download mutation indexes into the Library?">
            </param>
            <when value="true">
                <param name="source_url" type="select" label="Select a File"
                    dynamic_options="get_mutation_index_urls()" 
                    help="Select CTAT Mutation Indexes 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?" />
            </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 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.
        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 indexes included. Those must be downloaded
        separately. If you are not going to be using the mutation tool, uncheck the Download mutation indexes check box and
        save the space and time it takes to include the mutation index files. 
        - FIX - 
        This version of the tool does not yet implement the download of mutation indexes.
        - FIX -
        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 http="https://github.com/FusionFilter/FusionFilter/wiki">FusionFilter</a>
    </help>
    <code file="add_ctat_resource_lib.py" />
</tool>