view ena_consensus_submit.xml @ 3:7d751b5943b0 draft default tip

Uploaded
author ieguinoa
date Tue, 22 Feb 2022 11:03:34 +0000
parents 1ecd8ce07db4
children
line wrap: on
line source

<tool id="ena_consensus_submit" name="Submit consensus sequence to ENA" version="@VERSION@" python_template_version="3.5">
    <macros>
        <token name="@VERSION@">4.3.0</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[
#if $test_submit == "True":
    #set $credentials='./test_fake_path';
    touch $credentials;
    echo "username:test_user" >> $credentials;
    echo "password:test_password" >> $credentials;
#end if

webin_id=`grep 'username' $credentials | cut -d':' -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` | sed -e 's/^[[:space:]]*//'`;
      password=`grep 'password' \$GALAXY_ENA_SECRETS | cut -d':' -f2,2` | sed -e 's/^[[:space:]]*//'`;
      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 -d':' -f2,2`;
fi;


## create a manifest 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;
#if $min_gap_length:
    echo -e 'MINGAPLENGTH\t$min_gap_length' >> $manifest_base;
#end if
echo -e 'MOLECULETYPE\t$molecule_type' >> $manifest_base;
        
#if $metadata_file_or_form.metadata_format == "file":
    #import re
    #for $file in $metadata_file_or_form.genome_fasta:
        #if $file.is_of_type('fasta'):
            #set $full_name = $file.element_identifier + '.gz'
            gzip -c $file > './fasta/$full_name';
        #else:
            ln -s $file './fasta/$file.element_identifier';
        #end if

    #end for
    ## 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_files './manifests' $manifest_base >> $webin_cli_log;
    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;
    #set $study_id = $metadata_file_or_form.study_accession
    #set $sample_id = $metadata_file_or_form.sample_accession
    echo -e 'STUDY\t$study_id' >> $generated_manifest;
    echo -e 'SAMPLE\t$sample_id' >> $generated_manifest;
    center_name='$metadata_file_or_form.center_name';
    echo -e 'NAME\t$metadata_file_or_form.assembly_name' >> $generated_manifest;
    echo -e 'PLATFORM\t$metadata_file_or_form.sequencing_platform' >> $generated_manifest;
    #if $metadata_file_or_form.genome_fasta.is_of_type('fasta'):
        gzip -c $metadata_file_or_form.genome_fasta > consensus.fasta.gz;
    #else:
        ln -s $metadata_file_or_form.genome_fasta consensus.fasta.gz;
    #end if
    echo -e 'FASTA\tconsensus.fasta.gz' >> $generated_manifest;
#end if

#set $outputs_dir = 'outputs'
mkdir $outputs_dir;
#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`;
        echo "Submitting manifest \$manifest" >> $webin_cli_log;
        ena-webin-cli
        -context genome
        -userName "'\$webin_id'"
        -password "'\$password'"
        -centerName "'\$center_name'" 
        -manifest \$manifest
        -inputDir './fasta'
        #if $dry_run == "true":
            -validate
        #end if
        -outputDir $outputs_dir
        >> $webin_cli_log;
    done < submit_list.tab;

#else:
    ena-webin-cli
    #if $submit_test == "true":
        -test
    #end if
    #if $dry_run == "true":
        -validate
    #else:
        -submit
    #end if
    -context genome 
    -manifest $generated_manifest
    -password "'\$password'"
    -userName "'\$webin_id'"
    -inputDir "./"
    -centerName "'\$center_name'"
    -outputDir $outputs_dir
    >> $webin_cli_log ;
    ##cp ./genome/$metadata_file_or_form.assembly_name/validate/webin-cli.report $validate_output
#end if
tar -cf $webin_cli_outputs $outputs_dir ;
]]></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 $webin_id != "":
    username:$webin_id
    password:$password
#end if

        ]]></configfile>
    <configfile name="genome_fasta_files">
#import json
#import re
#if $metadata_file_or_form.metadata_format == "file":
    #set $fasta_files_list = list()
    #for $file in $metadata_file_or_form.genome_fasta:
        $fasta_files_list.append(str($file.element_identifier))
    #end for
    #echo json.dumps($fasta_files_list)
#end if
        </configfile>
    </configfiles>
<inputs>
    <param name="test_submit" type="hidden" value="False" />
    <param name="submit_test" type="boolean" truevalue="true" falsevalue="false" label="Submit to test server" help="use Webin test service instead of the production service. Please note that the Webin upload area is shared between test and production services, and that test submission files will not be archived." />
    <param name="dry_run" type="boolean" truevalue="true" falsevalue="false" label="Validate files and metadata but do not submit" help="Generate input files and run Webin-CLI with -validate option. If 'No' is selected then it will validate and submit (-submit flag)"/>
    <param name="assembly_type" type="select" label="Assembly type">
        <option value="clone">Clone</option>
        <option value="isolate">Isolate</option>
        <option value="COVID-19 outbreak">COVID-19 outbreak</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="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="form" selected="True">Fill in required submission metadata</option>
            <option value="file">(Beta) I used the Galaxy ENA upload tool to submit the raw data: parse the received submission receipt</option>
        </param>
        <when value="file">
            <param type="data" format="txt" name="ena_receipt" label="Submission receipt obtained from ENA upload tool"/>
            <param name="genome_fasta" type="data" label="Select the consensus sequence assembly files or a collection of them" format="fasta,fasta.gz" multiple="true"/>
        </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"/>
            <param name="sample_accession" type="text" optional="False" label="Sample accession"/>
            <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"/>
            <param name="genome_fasta" type="data" label="Select the consensus sequence assembly file" format="fasta,fasta.gz"/>
        </when>
    </conditional>
    <param name="min_gap_length" type="text" optional="True" label="Minimum gap length (optional)"/>
    </inputs>
    <outputs>
        <collection name="generated_manifests" type="list" label="Generated manifests">
            <discover_datasets pattern="(?P&lt;designation&gt;.+)\.txt" ext="txt" directory="manifests/"/>
        </collection>
        <data name="webin_cli_log" label="ENA submission log" format="txt"/>
        <data name="webin_cli_outputs" label="Webin cli outputs" format="tar"/>
        
    </outputs>
    <tests>
        <test>
            <param name="submit_test" value="true" />
            <param name="dry_run" value="true" />
            <param name="test_submit" value="True" />
            <param name="assembly_type" value="isolate"/>
            <param name="assembly_program" value="Test assembly program"/>
            <param name="molecule_type" value="viral cRNA"/>
            <param name="coverage" value="10000"/>
            <conditional name="metadata_file_or_form">
                <param name="metadata_format" value="form"/>
                <param name="assembly_name" value="Test assembly name"/>
                <param name="study_accession" value="PRJEB49173"/>
                <param name="sample_accession" value="SAMEA11953908"/>
                <param name="sequencing_platform" value="Nanopore 0011"/>
                <param name="description" value="Test Description"/>
                <param name="center_name" value="Test center name"/>
                <param name="genome_fasta" value="phiX2.fasta"/>
            </conditional>
            <param name="min_gap_length" value="30"/>
            <output name="webin_cli_log">
                <assert_contents>
                    <has_n_lines n="4"/>
                    <has_text_matching expression="ERROR: Invalid submission account user name or password."/>
                </assert_contents>
            </output>
        </test>
        <test>
            <param name="submit_test" value="true" />
            <param name="dry_run" value="true" />
            <param name="test_submit" value="True" />
            <param name="assembly_type" value="isolate"/>
            <param name="assembly_program" value="Test assembly program"/>
            <param name="molecule_type" value="viral cRNA"/>
            <param name="coverage" value="10000"/>
            <conditional name="metadata_file_or_form">
                <param name="metadata_format" value="file"/>
                <param name="ena_receipt" value="receipt_sample_noPhiX.txt"/>
                <param name="genome_fasta" value="phiX2.fasta.gz,sample_alias_001.fasta.gz"/>
            </conditional>
            <param name="min_gap_length" value="30"/>
            <output name="webin_cli_log">
                <assert_contents>
                    <has_text_matching expression="Processing phiX2"/>
                    <has_text_matching expression="No metadata found for sample phiX2"/>
                    <has_text_matching expression="Processing sample_alias_001"/>
                    <has_text_matching expression="Submitting manifest ./manifests/sample_alias_001.manifest.txt"/>
                    <has_text_matching expression="ERROR: Invalid submission account user name or password. Please try enclosing your password in single quotes."/>
                </assert_contents>
            </output>
        </test>
        <test>
            <param name="submit_test" value="true" />
            <param name="dry_run" value="true" />
            <param name="test_submit" value="True" />
            <param name="assembly_type" value="isolate"/>
            <param name="assembly_program" value="Test assembly program"/>
            <param name="molecule_type" value="viral cRNA"/>
            <param name="coverage" value="10000"/>
            <conditional name="metadata_file_or_form">
                <param name="metadata_format" value="file"/>
                <param name="ena_receipt" value="receipt_sample.txt"/>
                <param name="genome_fasta" value="sample_alias_001.fasta.gz"/>
            </conditional>
            <param name="min_gap_length" value="30"/>
            <output name="webin_cli_log">
                <assert_contents>
                    <has_text_matching expression="ERROR: Invalid submission account user name or password. Please try enclosing your password in single quotes."/>
                </assert_contents>
            </output>
        </test>
    </tests>
    <help><![CDATA[
        This tool is a wrapper for the ENA Webin CLI submission tool (https://ena-docs.readthedocs.io/en/latest/submit/general-guide/webin-cli.html).
    ]]></help>
</tool>