view ena_consensus_submit.xml @ 0:e25357392813 draft

Uploaded
author ieguinoa
date Tue, 18 May 2021 16:30:52 +0000
parents
children f24eb2f2cb0c
line wrap: on
line source

<tool id="ena_consensus_submit" name="Submit consensus sequence to ENA" version="0.1.0" python_template_version="3.5">
    <macros>
        <token name="@VERSION@">3.7.0</token>
    </macros>
    <requirements>
        <requirement type="package" version="@VERSION@">ena-webin-cli</requirement>
        <requirement type="package" version="1.76">biopython</requirement>
        <requirement type="package" version="5.3">pyyaml</requirement>
    </requirements>
    <command detect_errors="exit_code"><![CDATA[

webin_id=`grep 'username' $credentials | cut -f2,2`;
if [ "\$webin_id" = "" ]; then
  ## No credentials in user defined preferences    
  ## Fallback to global defined credentials (if exist)   
  #import os
  #if os.path.isfile(os.environ.get('GALAXY_ENA_SECRETS', '')):
      credentials_path=\${GALAXY_ENA_SECRETS};     
      webin_id=`grep 'username' \$GALAXY_ENA_SECRETS | cut -d' ' -f2,2`;
      password=`grep 'password' \$GALAXY_ENA_SECRETS | cut -d' ' -f2,2`;
      if [ "\$webin_id" = "" ]; then
          echo "No global credentials defined. Check your GALAXY_ENA_SECRETS file or set your credentials via: User -> Preferences -> Manage Information";
          exit 1;
      fi;
  #else:
      echo "No ENA credentials defined. Set your credentials via: User -> Preferences -> Manage Information";
      exit 1;
  #end if
else
  password=`grep 'password' $credentials | cut -f2,2`;
fi;

## if it is still running at this point then the webin_id and password are set

## create a manifaste base file with the parameters that are identical for all sequences that will be submited (if submitting a multifasta)
#set $manifest_base = 'manifest_base.tab'
mkdir manifests;
mkdir fasta;

echo -e "ASSEMBLY_TYPE\t$assembly_type" >> $manifest_base;
echo -e "COVERAGE\t$coverage" >> $manifest_base;
echo -e "PROGRAM\t$assembly_program" >> $manifest_base;
echo -e "MINGAPLENGTH\t$min_gap_length" >> $manifest_base;
echo -e "MOLECULETYPE\t$molecule_type" >> $manifest_base;
        
#if $metadata_file_or_form.metadata_format == "file":
    ## process the input tables, this creates an intermediate file with information
    python3 '$__tool_directory__/process_input.py' $metadata_file_or_form.ena_receipt $genome_fasta './manifests' './fasta' $manifest_base;
    center_name=`grep 'center_name' $metadata_file_or_form.ena_receipt | cut -f2,2 | tr -d '\n'`;
#else:
    #set $generated_manifest='./manifests/generated_manifest.txt'
    cp $manifest_base $generated_manifest
    $study_id = $metadata_file_or_form.study_accession
    $sample_id = $metadata_file_or_form.sample_accession
    echo "STUDY\t$study_id" > $generated_manifest;
    echo "SAMPLE\t$sample_id" >> $generated_manifest;
    center_name=$metadata_file_or_form.center_name;
    echo "ASSEMBLY_NAME\t$metadata_file_or_form.assembly_name" >> $generated_manifest;
    echo "PLATFORM\t$platform_name" >> $generated_manifest;
#end if
        

#if $metadata_file_or_form.metadata_format == "file":
    ## iterate over the list of manifest - fasta generated by the process_input
    ## in case of errors, this list is empty
    while read line; do
        manifest=`echo \$line | cut -d' ' -f1,1`;
        ena-webin-cli
        -context genome
        -userName \$webin_id
        -password \$password
        -centerName \$center_name 
        -manifest \$manifest
        -inputDir "./fasta"
        #if $dry_run == "true":
            -validate
        #end if
        ;
    done < submit_list.tab

#else:
    gzip -c $genome_fasta > consensus.fasta.gz;
    echo -e "FASTA\tconsensus.fasta.gz" >> $generated_manifest;
    ena-webin-cli
    -context genome 
    -userName \$webin_id
    -password \$password
    -centerName '\$center_name' 
    -manifest $generated_manifest
    -inputDir "./"
    #if $submit_test == "true":
        -test
    #end if
    #if $dry_run == "true":
        -validate
    #else:
        -submit
    #end if
#end if
]]></command>
    <configfiles>
        <configfile name="credentials"><![CDATA[
#set $webin_id = $__user__.extra_preferences.get('ena_webin_account|webin_id', "").strip()
#set $password = $__user__.extra_preferences.get('ena_webin_account|password', "").strip()
#if $webin_id != "":
    username\t"$webin_id"
    password\t"$password"
#end if
        ]]></configfile>
    </configfiles>
<inputs>
    <param name="submit_test" type="boolean" default="False" label="Submit to test server" help="Suggested to test metadata format" />
    <param name="dry_run" type="boolean" default="False" label="Validate files and metadata but do not submit" help="Generate input files and run Webin-CLI with -validate option."/>
    <param name="genome_fasta" type="data" label="Select the consensus sequence assembly file" format="fasta"/>
    <param name="assembly_type" type="select" label="Assembly type">
        <option value="clone">Clone</option>
        <option value="isolate">Isolate</option>
    </param>
    <param name="assembly_program" type="text" optional="False" label="Assembly program"/>
    <param name="molecule_type" type="select" label="Molecule type">
        <option value="genomic RNA" selected="True">genomic RNA</option>
        <option value="viral cRNA">viral cRNA</option>
        <option value="genomic DNA">genomic DNA</option>
    </param>
    <param name="min_gap_length" type="text" optional="True" label="Minimum gap length"/>
    <param name="coverage" type="float" optional="False" value="10000" label="Coverage"/>
    <conditional name="metadata_file_or_form">
        <param name="metadata_format" type="select" label="Select the method to load study and sample metadata">
            <option value="file" selected="True">I used Galaxy ENA upload tool for raw data submission, parse my submission receipt</option>
            <option value="form">Fill in required metadata for linking</option>
        </param>
        <when value="file">
            <param type="data" format="txt" name="ena_receipt" label="Submission receipt obtained from ENA upload tool"/>
        </when>
        <when value="form">
            <param name="assembly_name" type="text" optional="False" label="Assembly name"/>
            <param name="study_accession" type="text" optional="False" label="Study accession or unique name (alias)"/>
            <param name="sample_accession" type="text" optional="False" label="Sample accession or unique name (alias)"/>
            <param name="sequencing_platform" type="text" optional="False" label="Sequencing platform"/>
            <param name="description" type="text" optional="True" value="" label="Description" help="Free text description of the genome assembly (optional)"/>
            <param name="center_name" type="text" optional="False" label="Center name"/>
        </when>
    </conditional>
    </inputs>
    <outputs>
        <collection name="generated_manifests" type="list" label="Generated manifests">
            <!--<discover_datasets pattern="__name_and_ext__" directory="manifests" />-->
            <discover_datasets pattern="(?P&lt;designation&gt;.+)\.manifest\.txt$" ext="txt" directory="manifests" />
        </collection>
        <collection name="manifests_reports" type="list" label="Manifests reports">
            <!--[><discover_datasets pattern="__name_and_ext__" directory="manifests" /><]-->
            <discover_datasets pattern="(?P&lt;designation&gt;.+)\.manifest\.txt\.report$" ext="txt" directory="manifests" />
        </collection>
        <data name="webin_cli_report" label="ENA submission receipt" format="txt" from_work_dir="manifests/webin-cli.report"/>
    </outputs>
    <help><![CDATA[
        TODO: Fill in help.
    ]]></help>
</tool>