annotate add_protein_features_mouse.xml @ 0:1e856941a888 draft

planemo upload commit 8de59a5dbf9206fa3d9f7a1c07e79ccb792b3e3f-dirty

<tool id="prot_features_mouse" name="Add protein features (Mouse)" version="2019.02.08">
<description>[UniProt]
</description>
<requirements>
  <requirement type="package" version="3.4.1">R</requirement>
</requirements>
<stdio>
  <exit_code range="1:" />
</stdio>
<command><![CDATA[

  Rscript $__tool_directory__/add_protein_features_mouse.R 
  --inputtype="$inputtype.filetype"
  --input='$inputtype.genelist'

  #if $inputtype.filetype == "file" 
    --column='$inputtype.column' 
    --header=$inputtype.header
  #end if

  --pc_features='$pc_features'   
  --output='$output'  
  --uniprot_file=$__tool_directory__/tool-data/uniprot_features_mouse.tsv
    
]]></command>

<inputs>
  <conditional name="inputtype">
    <param name="filetype" type="select" label="Enter your IDs (UniProt-AC)" help="Copy/paste or from a file" > 
      <option value="file" selected="true">Input file containing your IDs </option>
      <option value="copy_paste">Copy/paste your list of IDs</option> 
    </param>
    <when value="copy_paste">
      <param name="genelist" type="text" label="Enter a list of IDs separated by tab, space or carriage return into the form field" help="">
        <sanitizer invalid_char="">
            <valid initial="string.printable">
                <remove value="&apos;"/>
            </valid>
            <mapping initial="none">
                <add source="&apos;" target="__sq__"/>
                <add source="&#x20;" target=""/>
                <add source="&#xA;" target=""/>
                <add source="&#xD;" target=""/>
                <add source="&#x9;" target=""/>
            </mapping>
        </sanitizer>
      </param>
    </when>
    <when value="file">
      <param name="genelist" type="data" format="txt,tabular" label="Select your file" help=""/>
      <param name="column" type="text" label="Column IDs (e.g : Enter c1 for column n°1)" value="c1"/> 
      <param name="header" type="boolean" checked="true" truevalue="true" falsevalue="false" label="Does input file have header?" />

    </when>
  </conditional>
    <param name="pc_features" type="select" label="Features/Annotation" multiple="true" help="" display="checkboxes" optional="true">
      <option value="Entry name" selected="false">Entry name (e.g. PLK3_MOUSE)</option>
      <option value="Protein names" selected="false">Protein names (e.g. Serine/threonine-protein kinase PLK3 (EC 2.7.11.21))</option>
      <option value="Length" selected="false">Sequence length</option>
      <option value="Mass" selected="false">Molecular Mass (Da)</option>
      <option value="Features" selected="false">Features (e.g. Chain (1); Transmembrane (4))</option>
      <option value="Intramembrane" selected="false">Intramembrane (e.g. INTRAMEM 104 116 Helical; Name=Pore helix 1. {ECO:0000250|UniProtKB:O00180})</option>
      <option value="Transmembrane" selected="false">Transmembrane (e.g. TRANSMEM 1 21 Helical. {ECO:0000255})</option>
      <option value="Topological domain" selected="false">Topological domain (e.g. TOPO_DOM 33 64 Extracellular. {ECO:0000250|UniProtKB:P11169})</option>
      <option value="Pathway" selected="false">Pathway (e.g. Steroid metabolism; cholesterol metabolism)</option>
      <option value="Function [CC]" selected="false">Function (e.g. FUNCTION: Serine/threonine-protein kinase involved in cell cycle regulation)</option>
      <option value="Post-translational modification" selected="false">Post-translational modification (e.g. PTM: Phosphorylation on Ser-300 or Ser-302 and on Ser-420)</option>
      <option value="Subcellular location [CC]" selected="false">Subcellular location (e.g. SUBCELLULAR LOCATION: Secreted)</option>
      <option value="Subunit structure [CC]" selected="false">Subunit structure (e.g. SUBUNIT: Homodimer. Heterodimerizes with YWHAE (By similarity))</option>
      <option value="Domain [CC]" selected="false">Domain (e.g. DOMAIN: The Walker motif (consensus sequence G-X-X-G-X-G-K-[ST]-T) is expected to bind ATP)</option>
      <option value="Tissue specificity" selected="false">Tissue specificity (e.g. TISSUE SPECIFICITY: Highest expression in fat cells {ECO:0000269|PubMed:7629182})</option>
      <option value="Involvement in disease" selected="false">Involvement in disease (e.g. DISEASE: Note=Defects in Btk are the cause of murine X-linked immunodeficiency (XID))</option>     
    </param>
</inputs>

<outputs>
  <data name="output" format="tsv" label="Add_information_from_UniProt on ${inputtype.genelist.name}">
    <filter>inputtype=="file"</filter>
  </data>
  <data name="output" format="tsv" label="Add_information_from_UniProt"/>
</outputs>

<tests>
  <test>
    <conditional name="inputtype">
      <param name="filetype " value="file"/>
      <param name="genelist" value="Wilson-foie-souris-up.txt"/>
      <param name="column" value="c1"/>
      <param name="header" value="true"/>
    </conditional>

    <param name="pc_features" value='Entry name,Protein names,Length,Mass,Features,Intramembrane,Transmembrane,Topological domain,Pathway,Function [CC],Post-translational modification,Subcellular location [CC],Subunit structure [CC],Domain [CC],Tissue specificity,Involvement in disease'/> 
      
    <output name="output" file="results_wilson-foie-souris-up.tsv"/>
  </test>
</tests>

<help><![CDATA[

**Description**

This tool retrieves annotation (protein features) from the UniProt database (Mouse proteins) to enrich your protein IDs list.

-----

**Input**

A list of Uniprot Accession Number (e.g. P05090) entered in a copy/paste mode or a file containing one or multiple columns with **at least one column of Uniprot accession number**. If your input file contains other type of IDs, please use the ID_Converter tool.  

-----

**Parameters**

"Select features": three categories of annotation can be retrieved: physico-chemical features, localisation (chromosome, subcellular) and disease information (set to "Yes" by default). Select each feature according to your interest by clicking the corresponding checkbox. 

You can get more information about annotations `here <https://www.uniprot.org/help/uniprotkb_column_names>`_.


-----

**Output**

Output is a tabular file containing both original columns and new columns including the annotation requested.  
Please, note that a "NA" is returned when there is no information available from UniProt.

-----

**Data source (release date)**

Annotations have been retrieved from UniProt.org (02/2019).

-----

.. 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, FR

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>