annotate seurat_macros.xml @ 1:834f7d8e5b4c draft default tip

planemo upload commit 0264c359f1d638bbbbab515a3502231f679cdcf6

<?xml version="1.0"?>
<macros>
    <token name="@VERSION@">4.0.0</token>
    <token name="@SEURAT_VERSION@">4.0.4</token>
    <xml name="requirements">
        <requirements>
            <requirement type="package" version="@VERSION@">seurat-scripts</requirement>
        </requirements>
    </xml>
    <xml name="version">
    	<version_command><![CDATA[
echo $(R --version | grep version | grep -v GNU)", seurat version" $(R --vanilla --slave -e "library(seurat); cat(sessionInfo()\$otherPkgs\$seurat\$Version)" 2> /dev/null | grep -v -i "WARNING: ")
    ]]></version_command>
    </xml>

   <xml name="input_object_params" token_multiple="False" token_varname="input" token_optional="False">
     <conditional name="@VARNAME@" label="Input format">
       <param type="select" name="format" label="Choose the format of the @VARNAME@" help="Seurat RDS, Seurat H5, Single Cell Experiment RDS, Loom or AnnData">
         <option value="rds_seurat" selected="true">RDS with a Seurat object</option>
         <option value="loom">Loom</option>
         <option value="h5seurat">Seurat HDF5</option>
         <option value="anndata">AnnData</option>
         <option value="rds_sce">RDS with a Single Cell Experiment object</option>
       </param>
       <when value="anndata">
         <param type="data" name="anndata_file" multiple="@MULTIPLE@" optional="@OPTIONAL@" label="AnnData file" help="Select AnnData files for @VARNAME@" format="h5,h5ad"/>
       </when>
       <when value="loom">
         <param type="data" name="loom_file" multiple="@MULTIPLE@" optional="@OPTIONAL@" label="Loom file" help="Select Loom file(s) for @VARNAME@" format="h5,h5loom"/>
       </when>
       <when value="rds_seurat">
         <param type="data" name="rds_seurat_file" multiple="@MULTIPLE@" optional="@OPTIONAL@" label="RDS file" help="Select RDS file(s) with Seurat object for @VARNAME@" format="rdata"/>
       </when>
       <when value="rds_sce">
         <param type="data" name="rds_sce_file" multiple="@MULTIPLE@" optional="@OPTIONAL@" label="RDS file" help="Select RDS file(s) with Single Cell Experiment object for @VARNAME@" format="rdata"/>
       </when>
       <when value="h5seurat">
         <param type="data" name="h5seurat_file" multiple="@MULTIPLE@" optional="@OPTIONAL@" label="Seurat HDF5" help="Select Seurat HDF5 file(s) for @VARNAME" format="h5"/>
       </when>
     </conditional>
   </xml>

   <token name="@INPUT_OBJ_PREAMBLE@">
     #if $input.format == 'loom'
     ln -s '$input.loom_file' input.loom;
     #else if $input.format == 'h5seurat'
     ln -s '$input.h5seurat_file' input.h5seurat;
     #else if $input.format == 'anndata'
     ## it complains when using links for AnnData...
     cp '$input.anndata_file' input.h5ad;
     #end if
   </token>

   <token name="@INPUT_OBJECT@">
    #if $input.format == "anndata"
        --input-object-file input.h5ad --input-format anndata
    #else if $input.format == "loom"
        --input-object-file input.loom --input-format loom
    #else if $input.format == "rds_seurat"
        --input-object-file '$input.rds_seurat_file' --input-format seurat
    #else if $input.format == "rds_sce"
        --input-object-file '$input.rds_sce_file' --input-format singlecellexperiment
    #else if $input.format == "h5seurat"
        --input-object-file input.h5seurat --input-format h5seurat
    #end if
   </token>

   <token name="@QUERY_OBJ_PREAMBLE@">
     #if $query.format == 'loom'
     ln -s '$query.loom_file' query.loom;
     #else if $query.format == 'h5seurat'
     ln -s '$query.h5seurat_file' query.h5seurat;
     #else if $query.format == 'anndata'
     ## it complains when using links for AnnData...
     cp '$query.anndata_file' query.h5ad;
     #end if
   </token>

   <token name="@QUERY_OBJECT@">
    #if $query.format == "anndata"
        --query-object-file query.h5ad --query-format anndata
    #else if $query.format == "loom"
        --query-object-file query.loom --query-format loom
    #else if $query.format == "rds_seurat"
        --query-object-file '$query.rds_seurat_file' --query-format seurat
    #else if $query.format == "rds_sce"
        --query-object-file '$query.rds_sce_file' --query-format singlecellexperiment
    #else if $query.format == "h5seurat"
        --query-object-file query.h5seurat --query-format h5seurat
    #end if
  </token>

  <token name="@ANCHORS_OBJ_PREAMBLE@">
    #if $anchors.format == 'loom'
    ln -s '$anchors.loom_file' anchors.loom;
    #else if $anchors.format == 'h5seurat'
    ln -s '$anchors.h5seurat_file' anchors.h5seurat;
    #else if $anchors.format == 'anndata'
    ## it complains when using links for AnnData...
    cp '$anchors.anndata_file' anchors.h5ad;
    #end if
  </token>

  <token name="@ANCHORS_OBJECT@">
   #if $anchors.format == "anndata"
       --anchors-object-file anchors.h5ad --anchors-format anndata
   #else if $anchors.format == "loom"
       --anchors-object-file anchors.loom --anchors-format loom
   #else if $anchors.format == "rds_seurat"
       --anchors-object-file '$anchors.rds_seurat_file' --anchors-format seurat
   #else if $anchors.format == "rds_sce"
       --anchors-object-file '$anchors.rds_sce_file' --anchors-format singlecellexperiment
   #else if $anchors.format == "h5seurat"
       --anchors-object-file anchors.h5seurat --anchors-format h5seurat
   #end if
 </token>

 <token name="@REFERENCE_OBJ_PREAMBLE@">
   #if $reference.format == 'loom'
   ln -s '$reference.loom_file' reference.loom;
   #else if $reference.format == 'h5seurat'
   ln -s '$reference.h5seurat_file' reference.h5seurat;
   #else if $reference.format == 'anndata'
   ## it complains when using links for AnnData...
   cp '$reference.anndata_file' reference.h5ad;
   #end if
 </token>

 <token name="@REFERENCE_OBJECT@">
  #if $reference.format == "anndata"
      --reference-object-file reference.h5ad --reference-format anndata
  #else if $reference.format == "loom"
      --reference-object-file reference.loom --reference-format loom
  #else if $reference.format == "rds_seurat"
      --reference-object-file '$reference.rds_seurat_file' --reference-format seurat
  #else if $reference.format == "rds_sce"
      --reference-object-file '$reference.rds_sce_file' --reference-format singlecellexperiment
  #else if $reference.format == "h5seurat"
      --reference-object-file reference.h5seurat --reference-format h5seurat
  #end if
 </token>

   <token name="@INPUT_OBJS_PREAMBLE@">
     #if $input.format == 'loom'
     #for $i, $fh in enumerate($input.loom_file):
        ln -s '$fh' input.${i}.loom;
     #end for
     #else if $input.format == 'h5seurat'
     #for $i, $fh in enumerate($input.h5seurat_file):
        ln -s '$fh' input.${i}.h5seurat;
     #end for
     #else if $input.format == 'anndata'
     ## it complains when using links for AnnData...
     #for $i, $fh in enumerate($input.anndata_file):
        cp '$fh' input.${i}.h5ad;
     #end for
     #end if
   </token>

   <token name="@INPUT_OBJECTS@">
    #if $input.format == "anndata"
        --input-object-files
    #set file_array = [ "input."+str($i)+".h5ad" for $i, $fh in enumerate($input.anndata_file)]
    #set files = ",".join($file_array)
        ${files}
         --input-format anndata
    #else if $input.format == "loom"
        --input-object-files
    #set file_array = [ "input."+str($i)+".loom" for $i, $fh in enumerate($input.loom_file)]
    #set files = ",".join($file_array)
        ${files}
         --input-format loom
    #else if $input.format == "rds_seurat"
        --input-object-files
    #set file_array = $input.rds_seurat_file
    #set files = ",".join([ str($fh) for $fh in $file_array ])
        ${files}
        --input-format seurat
    #else if $input.format == "rds_sce"
        --input-object-files
    #set file_array = $input.rds_sce_file
    #set files = ",".join([ str($fh) for $fh in $file_array ])
        ${files}
        --input-format singlecellexperiment
    #else if $input.format == "h5seurat"
        --input-object-files
    #set file_array = [ "input."+str($i)+".h5seurat" for $i, $fh in enumerate($input.h5seurat)]
    #set files = ",".join($file_array)
        ${files}
        --input-format h5seurat
    #end if
   </token>

   <token name="@REFERENCE_OBJS_PREAMBLE@">
     #if $reference.format == 'loom'
     #for $i, $fh in enumerate($reference.loom_file):
        ln -s '$fh' reference.${i}.loom;
     #end for
     #else if $reference.format == 'h5seurat'
     #for $i, $fh in enumerate($reference.h5seurat_file):
        ln -s '$fh' reference.${i}.h5seurat;
     #end for
     #else if $reference.format == 'anndata'
     ## it complains when using links for AnnData...
     #for $i, $fh in enumerate($reference.anndata_file):
        cp '$fh' reference.${i}.h5ad;
     #end for
     #end if
   </token>

   <token name="@REFERENCE_OBJECTS@">
    #if $reference.format == "anndata" and $reference.anndata_file:
        --reference-object-files
    #set file_array = [ "reference."+str($i)+".h5ad" for $i, $fh in enumerate($reference.anndata_file)]
    #set files = ",".join($file_array)
        ${files}
         --reference-format anndata
    #else if $reference.format == "loom" and $reference.loom_file:
        --reference-object-files
    #set file_array = [ "reference."+str($i)+".loom" for $i, $fh in enumerate($reference.loom_file)]
    #set files = ",".join($file_array)
        ${files}
         --reference-format loom
    #else if $reference.format == "rds_seurat" and $reference.rds_seurat_file:
        --reference-object-files
    #set files = ",".join([ str($fh) for $fh in $reference.rds_seurat_file ])
        ${files}
        --reference-format seurat
    #else if $reference.format == "rds_sce" and $reference.rds_sce_file:
        --reference-object-files
    #set files = ",".join([ str($fh) for $fh in $reference.rds_sce_file ])
        ${files}
        --reference-format singlecellexperiment
    #else if $reference.format == "h5seurat" and $reference.h5seurat:
        --reference-object-files
    #set file_array = [ "reference."+str($i)+".h5seurat" for $i, $fh in enumerate($reference.h5seurat)]
    #set files = ",".join($file_array)
        ${files}
        --reference-format h5seurat
    #end if
   </token>

   <xml name="output_object_params">
     <param type="select" name="format" label="Choose the format of the output" help="Seurat, Single Cell Experiment, AnnData or Loom">
       <option value="rds_seurat" selected="true">RDS with a Seurat object</option>
       <option value="anndata">AnnData written by Seurat</option>
       <option value="loom">Loom</option>
       <option value="rds_sce">RDS with a Single Cell Experiment object</option>
     </param>
   </xml>

   <xml name="output_files">
    <data name="loom_file" from_work_dir="seurat_obj.loom" format="h5" label="${tool.name} on ${on_string}: Seurat Loom">
      <filter>format == 'loom'</filter>
    </data>
    <data name="rds_seurat_file" format="rdata" label="${tool.name} on ${on_string}: Seurat RDS">
      <filter>format == 'rds_seurat'</filter>
    </data>
    <data name="anndata_file" format="h5ad" label="${tool.name} on ${on_string}: AnnData from Seurat">
        <filter>format == 'anndata'</filter>
    </data>
    <data name="rds_sce_file" format="rdata" label="${tool.name} on ${on_string}: Seurat Single Cell Experiment RDS">
      <filter>format == 'rds_sce'</filter>
    </data>
   </xml>

   <token name="@OUTPUT_OBJECT@">
    #if $format == "anndata"
        --output-object-file '$anndata_file' --output-format anndata
    #else if $format == "loom"
        --output-object-file seurat_obj.loom --output-format loom
    #else if $format == "rds_seurat"
        --output-object-file '$rds_seurat_file' --output-format seurat
    #else if $format == "rds_sce"
        --output-object-file '$rds_sce_file' --output-format singlecellexperiment
    #end if
   </token>

   <xml name="plot_output_files_format" token_format="png">
     <data label="Seurat ${plot_type.plot_type_selector} on ${on_string}: @FORMAT@ plot" name="plot_out_@FORMAT@" format='@FORMAT@' >
       <filter>plot_format == '@FORMAT@'</filter>
     </data>
   </xml>

   <token name="@OUTPUT_PLOT@">
    #if $plot_format == "png"
        --plot-out '$plot_out_png'
    #else if $plot_format == "pdf"
        --plot-out '$plot_out_pdf'
    #else if $plot_format == "eps"
        --plot-out '$plot_out_eps'
    #else if $plot_format == "ps"
        --plot-out '$plot_out_ps'
    #else if $plot_format == "jpg"
        --plot-out '$plot_out_jpg'
    #else if $plot_format == "tiff"
        --plot-out '$plot_out_tiff'
    #else if $plot_format == "svg"
        --plot-out '$plot_out_svg'
    #end if
   </token>

    <xml name="genes-use-input">
      <param name="genes_use" argument="--genes-use" optional="true" type="data" format="tsv,txt,tabular" label="Genes to use" help="A file with gene names to use in construction of SNN graph if building directly based on expression data rather than a dimensionally reduced representation (i.e. PCs)."/>
    </xml>
    <xml name="dims-use-input">
      <param name="dims_use" argument="--dims-use" min="1" optional="true" type="integer" label="PCA Dimensions to use" help="Number of PCs (dimensions) to use in construction of the SNN graph."/>
    </xml>

    <token name="@SEURAT_INTRO@"><![CDATA[
Seurat_ is a toolkit for quality control, analysis, and exploration of single cell RNA sequencing data.
It is developed and maintained by the `Satija Lab`_ at NYGC. Seurat aims to enable users to identify and
interpret sources of heterogeneity from single cell transcriptomic measurements, and to integrate diverse
types of single cell data.
      ]]></token>

    <token name="@VERSION_HISTORY@"><![CDATA[
**Version history**
4.0.0: Moves to Seurat 4.0.0, introducing a number of methods for merging datasets, plus the whole suite of Seurat plots. Pablo Moreno with funding from AstraZeneca.

3.2.3+galaxy0: Moves to Seurat 3.2.3 and introduce convert method, improving format interconversion support.

3.1.2_0.0.8: Update metadata parsing

3.1.1_0.0.7: Exposes perplexity and enables tab input.

3.1.1_0.0.6+galaxy0: Moved to Seurat 3.

  Find clusters: removed dims-use, k-param, prune-snn.

2.3.1+galaxy0: Improved documentation and further exposition of all script's options. Pablo Moreno, Jonathan Manning and Ni Huang, Expression Atlas team https://www.ebi.ac.uk/gxa/home  at
EMBL-EBI https://www.ebi.ac.uk/. Parts obtained from wrappers from Christophe Antoniewski (GitHub drosofff) and Lea Bellenger (GitHub bellenger-l).

0.0.1: Initial contribution. Maria Doyle (GitHub mblue9).
      ]]></token>


    <xml name="citations">
        <citations>
            <citation type="doi">10.1038/s41592-021-01102-w</citation>
            <citation type="doi">10.1038/nbt.4096</citation>
            <citation type="bibtex">
              @misc{r-seurat-scripts.git,
              author = {Jonathan Manning, Pablo Moreno, EBI Gene Expression Team},
              year = {2018},
              title = {Seurat-scripts: command line interface for Seurat},
              publisher = {GitHub},
              journal = {GitHub repository},
              url = {https://github.com/ebi-gene-expression-group/r-seurat-scripts.git},
            }
            </citation>
        </citations>
    </xml>
</macros>