annotate data_manager/resource_building.xml @ 4:e967a99d66b3 draft

"planemo upload commit 540dd383c0617193db43bf11457011888751b022-dirty"

<tool id="data_manager_proteore" name="Get source files for proteore tools" version="2020.01.21" tool_type="manage_data">
<description>
to create or update reference files for proteore tools
</description>
<requirements>
    <!--requirement type="package" version="1.8.2">sparqlwrapper</requirement-->
</requirements>
<stdio>
  <exit_code range="1:" />
</stdio>
<command><![CDATA[

    python $__tool_directory__/resource_building.py
    #if $database.database == "human_protein_atlas"
        --hpa "$database.tissues"
    #else if $database.database == "peptide_atlas"
        --peptideatlas="$database.tissues"
        --date="$database.date"
    #else if $database.database == "id_mapping"
        --id_mapping="$database.species"
    #else if $database.database == "PPI"
        --interactome="$database.base.interactome"
        #if $database.base.interactome == "biogrid"
            --species="$database.base.species"
        #end if
    #else if $database.database == "nextprot"
        --database=$database.database
    #end if
    --output "$output"
    --tool_data_path=$__tool_data_path__
    
]]></command>

<inputs>
    <conditional name="database">
        <param name="database" type="select">
            <option value="human_protein_atlas">Human Protein Atlas</option>
            <option value="peptide_atlas">Peptide Atlas</option>
            <option value="id_mapping">ID mapping</option>
            <option value="PPI">Build protein interaction maps</option>
            <option value="nextprot">neXtProt</option>
        </param>
        <when value="human_protein_atlas">
            <param name="tissues" type="select" multiple="false" label="Please select tissue">
                <option value="HPA_normal_tissue">Normal tissue</option>
                <option value="HPA_pathology">Pathology</option>
                <option value="HPA_full_atlas">Full Atlas</option>
            </param>
        </when>
        <when value="peptide_atlas">
            <param name="tissues" type="select" multiple="false" label="Please select the tissue">
                <option value="432.Human_Adrenal_gland">Human Adrenal gland proteome</option>
                <option value="441.Human_Brain">Human Brain proteome</option>
                <option value="427.Human_Breast">Human Breast proteome</option>
                <option value="434.Human_CSF">Human CSF (Cerebro Spinal Fluid) proteome</option>
                <option value="374.Human_Colon_cancer">Human Colon cancer proteome</option>
                <option value="429.Human_Digestive_system">Human Digestive system proteome</option>
                <option value="430.Human_Female_reproductive_system">Human Female reproductive system proteome</option>
                <option value="418.Human_Heart">Human Heart proteome</option>
                <option value="424.Human_Kidney">Human Kidney proteome</option>
                <option value="425.Human_Liver">Human Liver proteome</option>
                <option value="419.Human_Lung">Human Lung proteome</option>
                <option value="431.Human_Male_reproductive_system">Human Male reproductive system proteome</option>
                <option value="420.Human_Pancreas">Human Pancreas proteome</option>
                <option value="465.Human_Plasma_non_glyco">Human Plasma non glyco proteome</option>
                <option value="421.Human_Spleen">Human Spleen proteome</option>
                <option value="463.Human_Testis">Human Testis proteome</option>
                <option value="422.Human_Urinary_bladder">Human Bladder proteome</option>
                <option value="423.Human_Urine">Human Urine proteome</option>
            </param>
            <param name="date" type="text" value="" label="enter the build date" help="for example: '2018-04'"/>
        </when>
        <when value="id_mapping">
            <param name="species" type="select" multiple="false" label="Please select the species">
                <option value="Human">Human (Homo sapiens)</option>
                <option value="Mouse">Mouse (Mus musculus)</option>
                <option value="Rat">Rat (Rattus norvegicus)</option>
            </param>
        </when>
        <when value="PPI">
            <conditional name="base">
                <param name="interactome" type="select" multiple="false" label="Please select interactome">
                    <option value="biogrid">BioGRID</option>
                    <option value="bioplex">Human Bioplex 2.0</option>
                    <option value="humap">Human protein complex Map (Hu.map)</option>
                </param>
                <when value="biogrid">
                    <param name="species" type="select" multiple="false" label="Please select the species">
                        <option value="Human">Human (Homo sapiens)</option>
                        <option value="Mouse">Mouse (Mus musculus)</option>
                        <option value="Rat">Rat (Rattus norvegicus)</option>
                    </param>
                </when>
                <when value="bioplex"/>
                <when value="humap"/>
            </conditional>
        </when>
        <when value="nextprot"/>
    </conditional>
</inputs>

<outputs>
    <!--data format="tabular" name="output">
        <discover_datasets pattern="(?P&lt;designation&gt;.+).tsv" ext="tabular" visible="true" assign_primary_output="true" />
    </data-->
    <data name="output" format="data_manager_json"/>
</outputs>

<tests>
</tests>

<help><![CDATA[

**Description**

This tool is a data manager designed to update resources files of ProteoRe tools. For now, only resources files for tools listed below are handled: 

* "Get MS/MS observations in tissue/fluid [Peptide Atlas]"
* "Get expression profiles by (normal or tumor) tissue/cell type [Human Protein Atlas]"
* "ID converter"

-----

**Input**

There's no input needed, once you selected the tool and file you want to update, it will be generated automatically.

-----

**Parameters**

* database: the database to update (for now one per tool)

Once a database is selected, there's a second dropdown menu to select the specific file you want to update.

* for 'Human Protein Atlas': 'Normal tissue', 'Pathology' and 'Full Atlas'

* for 'Peptide Atlas':  'Human liver', 'Human brain', 'Human heart', 'Human kidney', 'Human blood plasma', 'Human urine' and 'Human cerebrospinal fluid'

* for  'ID mapping': 'Human (Homo sapiens)', 'Mouse (Mus musculus)' and 'Rat (Rattus norvegicus)'

* for 'Build protein interaction maps': "BioGRID", "Bioplex" and "Human (Homo sapiens)", "Mouse (Mus musculus)", "Rat (Rattus norvegicus)"

-----

**Output**

The output is the reference file selected for update in input.

For example, if you select database="Human Protein Atlas" and Please select tissue="Normal tissue":

the output is a new reference file for "Get expression profiles by (normal or tumor) tissue/cell type [Human Protein Atlas]" 
dated from the day and listed in the dropdown menu "Normal tissue HPA version".

.. class:: warningmark

A reference file created with this data manager will appears in the concerned ProteoRE tool. It can not be removed with the data manager.

-----

**Data sources**

For 'Human Protein Atlas': 

* `Normal Tissue <https://www.proteinatlas.org/download/normal_tissue.tsv.zip>`_.
* `Pathology <https://www.proteinatlas.org/download/pathology.tsv.zip>`_.
* `Full Atlas <https://www.proteinatlas.org/download/proteinatlas.tsv.zip>`_.

For 'Peptide Atlas':


* `Human Adrenal gland proteome <https://db.systemsbiology.net/sbeams/cgi/PeptideAtlas/GetProteins?&atlas_build_id=432&display_options=ShowAbundances&organism_id=2&redundancy_constraint=4&presence_level_constraint=1%2C2&gene_annotation_level_constraint=leaf&QUERY_NAME=AT_GetProteins&action=QUERY&output_mode=tsv&apply_action=QUERY>`_.
* `Human Brain proteome <https://db.systemsbiology.net/sbeams/cgi/PeptideAtlas/GetProteins?&atlas_build_id=441&display_options=ShowAbundances&organism_id=2&redundancy_constraint=4&presence_level_constraint=1%2C2&gene_annotation_level_constraint=leaf&QUERY_NAME=AT_GetProteins&action=QUERY&output_mode=tsv&apply_action=QUERY>`_.
* `Human Breast Proteome <https://db.systemsbiology.net/sbeams/cgi/PeptideAtlas/GetProteins?&atlas_build_id=427&display_options=ShowAbundances&organism_id=2&redundancy_constraint=4&presence_level_constraint=1%2C2&gene_annotation_level_constraint=leaf&QUERY_NAME=AT_GetProteins&action=QUERY&output_mode=tsv&apply_action=QUERY>`_.
* `Human CSF proteome <https://db.systemsbiology.net/sbeams/cgi/PeptideAtlas/GetProteins?&atlas_build_id=434&display_options=ShowAbundances&organism_id=2&redundancy_constraint=4&presence_level_constraint=1%2C2&gene_annotation_level_constraint=leaf&QUERY_NAME=AT_GetProteins&action=QUERY&output_mode=tsv&apply_action=QUERY>`_.
* `Human Digestive System proteome <https://db.systemsbiology.net/sbeams/cgi/PeptideAtlas/GetProteins?&atlas_build_id=429&display_options=ShowAbundances&organism_id=2&redundancy_constraint=4&presence_level_constraint=1%2C2&gene_annotation_level_constraint=leaf&QUERY_NAME=AT_GetProteins&action=QUERY&output_mode=tsv&apply_action=QUERY>`_.
* `Human female reproductive system proteome <https://db.systemsbiology.net/sbeams/cgi/PeptideAtlas/GetProteins?&atlas_build_id=430&display_options=ShowAbundances&organism_id=2&redundancy_constraint=4&presence_level_constraint=1%2C2&gene_annotation_level_constraint=leaf&QUERY_NAME=AT_GetProteins&action=QUERY&output_mode=tsv&apply_action=QUERY>`_.
* `Human Heart proteome <https://db.systemsbiology.net/sbeams/cgi/PeptideAtlas/GetProteins?&atlas_build_id=418&display_options=ShowAbundances&organism_id=2&redundancy_constraint=4&presence_level_constraint=1%2C2&gene_annotation_level_constraint=leaf&QUERY_NAME=AT_GetProteins&action=QUERY&output_mode=tsv&apply_action=QUERY>`_.
* `Human Kidney man Kidney Proteome <https://db.systemsbiology.net/sbeams/cgi/PeptideAtlas/GetProteins?&atlas_build_id=424&display_options=ShowAbundances&organism_id=2&redundancy_constraint=4&presence_level_constraint=1%2C2&gene_annotation_level_constraint=leaf&QUERY_NAME=AT_GetProteins&action=QUERY&output_mode=tsv&apply_action=QUERY>`_.
* `Human Liver proteome <https://db.systemsbiology.net/sbeams/cgi/PeptideAtlas/GetProteins?&atlas_build_id=425&display_options=ShowAbundances&organism_id=2&redundancy_constraint=4&presence_level_constraint=1%2C2&gene_annotation_level_constraint=leaf&QUERY_NAME=AT_GetProteins&action=QUERY&output_mode=tsv&apply_action=QUERY>`_.
* `Human Lung proteome <https://db.systemsbiology.net/sbeams/cgi/PeptideAtlas/GetProteins?&atlas_build_id=419&display_options=ShowAbundances&organism_id=2&redundancy_constraint=4&presence_level_constraint=1%2C2&gene_annotation_level_constraint=leaf&QUERY_NAME=AT_GetProteins&action=QUERY&output_mode=tsv&apply_action=QUERY>`_.
* `Human Male Reproductive System proteome <https://db.systemsbiology.net/sbeams/cgi/PeptideAtlas/GetProteins?&atlas_build_id=431&display_options=ShowAbundances&organism_id=2&redundancy_constraint=4&presence_level_constraint=1%2C2&gene_annotation_level_constraint=leaf&QUERY_NAME=AT_GetProteins&action=QUERY&output_mode=tsv&apply_action=QUERY>`_.
* `Human Pancreas proteome <https://db.systemsbiology.net/sbeams/cgi/PeptideAtlas/GetProteins?&atlas_build_id=420&display_options=ShowAbundances&organism_id=2&redundancy_constraint=4&presence_level_constraint=1%2C2&gene_annotation_level_constraint=leaf&QUERY_NAME=AT_GetProteins&action=QUERY&output_mode=tsv&apply_action=QUERY>`_.
* `Human Plasma Non-Glyco proteome <https://db.systemsbiology.net/sbeams/cgi/PeptideAtlas/GetProteins?&atlas_build_id=465&display_options=ShowAbundances&organism_id=2&redundancy_constraint=4&presence_level_constraint=1%2C2&gene_annotation_level_constraint=leaf&QUERY_NAME=AT_GetProteins&action=QUERY&output_mode=tsv&apply_action=QUERY>`_.
* `Human Spleen proteome <https://db.systemsbiology.net/sbeams/cgi/PeptideAtlas/GetProteins?&atlas_build_id=421&display_options=ShowAbundances&organism_id=2&redundancy_constraint=4&presence_level_constraint=1%2C2&gene_annotation_level_constraint=leaf&QUERY_NAME=AT_GetProteins&action=QUERY&output_mode=tsv&apply_action=QUERY>`_.
* `Human Testis proteome <https://db.systemsbiology.net/sbeams/cgi/PeptideAtlas/GetProteins?&atlas_build_id=463&display_options=ShowAbundances&organism_id=2&redundancy_constraint=4&presence_level_constraint=1%2C2&gene_annotation_level_constraint=leaf&QUERY_NAME=AT_GetProteins&action=QUERY&output_mode=tsv&apply_action=QUERY>`_.
* `Human Urinary Bladder proteome <https://db.systemsbiology.net/sbeams/cgi/PeptideAtlas/GetProteins?&atlas_build_id=422&display_options=ShowAbundances&organism_id=2&redundancy_constraint=4&presence_level_constraint=1%2C2&gene_annotation_level_constraint=leaf&QUERY_NAME=AT_GetProteins&action=QUERY&output_mode=tsv&apply_action=QUERY>`_.
* `Human Urine proteome <https://db.systemsbiology.net/sbeams/cgi/PeptideAtlas/GetProteins?&atlas_build_id=423&display_options=ShowAbundances&organism_id=2&redundancy_constraint=4&presence_level_constraint=1%2C2&gene_annotation_level_constraint=leaf&QUERY_NAME=AT_GetProteins&action=QUERY&output_mode=tsv&apply_action=QUERY>`_.

For ID mapping:

* `HUMAN_9606_idmapping_selected.tab <ftp://ftp.uniprot.org/pub/databases/uniprot/current_release/knowledgebase/idmapping/by_organism/HUMAN_9606_idmapping_selected.tab.gz>`_.
* `HUMAN_9606_idmapping.dat <ftp://ftp.uniprot.org/pub/databases/uniprot/current_release/knowledgebase/idmapping/by_organism/HUMAN_9606_idmapping.dat.gz>`_.
* `nextprot_ac_list_all.txt <ftp://ftp.nextprot.org/pub/current_release/ac_lists/nextprot_ac_list_all.txt>`_.
* `MOUSE_10090_idmapping_selected.tab <ftp://ftp.uniprot.org/pub/databases/uniprot/current_release/knowledgebase/idmapping/by_organism/MOUSE_10090_idmapping_selected.tab.gz>`_.
* `MOUSE_10090_idmapping.dat <ftp://ftp.uniprot.org/pub/databases/uniprot/current_release/knowledgebase/idmapping/by_organism/MOUSE_10090_idmapping.dat.gz>`_.
* `RAT_10116_idmapping_selected.tab <ftp://ftp.uniprot.org/pub/databases/uniprot/current_release/knowledgebase/idmapping/by_organism/RAT_10116_idmapping_selected.tab.gz>`_.
* `RAT_10116_idmapping.dat <ftp://ftp.uniprot.org/pub/databases/uniprot/current_release/knowledgebase/idmapping/by_organism/RAT_10116_idmapping.dat.gz>`_.

For Build protein interaction maps:

* `BIOGRID_oragism <https://downloads.thebiogrid.org/Download/BioGRID/Release-Archive/BIOGRID-3.5.167/BIOGRID-ORGANISM-3.5.167.tab2.zip>`_.
* `NCBI2Reactome.txt <https://www.reactome.org/download/current/NCBI2Reactome.txt>`_.
* `Bioplex_interactionList_v4a.tsv <http://bioplex.hms.harvard.edu/data/BioPlex_interactionList_v4a.tsv>`_.
* `UniProt2Reactome.txt <https://reactome.org/download/current/UniProt2Reactome.txt>`_.

-----

.. class:: infomark

**Authors**

David Christiany, Lisa Peru, T.P. Lien Nguyen, Florence Combes, Yves Vandenbrouck CEA, INSERM, CNRS, Grenoble-Alpes University, BIG Institute, FR

Sandra Dérozier, Olivier Rué, Christophe Caron, Valentin Loux INRA, Paris-Saclay University, MAIAGE Unit, Migale Bioinformatics platform

This work has been partially funded through the French National Agency for Research (ANR) IFB project.

Contact [e-mail] for any questions or concerns about the Galaxy implementation of this tool.
 
    ]]></help>
    <citations>
    </citations>

</tool>