comparison icescreen.xml @ 9:cca2a89210ae draft default tip

planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/icescreen commit c45c31257dbfd6991c86cee56bb8961e4c5cbe47

8:130a217f8d3f

<tool id="icescreen" name="ICEscreen" version="@TOOL_VERSION@+galaxy@VERSION_SUFFIX@" profile="20.09">
    <description>detects and annotates ICEs (Integrative and Conjugative Elements) and IMEs (Integrative and Mobilizable Elements) in Bacillota genomes.</description>
    <macros>
        <token name="@TOOL_VERSION@">1.3.1</token>
        <token name="@VERSION_SUFFIX@">0</token>
    </macros>
    <xrefs>
        <xref type="bio.tools">icescreen</xref>
    </xrefs>
    <requirements>
        <requirement type="package" version="@TOOL_VERSION@">icescreen</requirement>
        <requirement type="package" version="3.0">zip</requirement>
    </requirements>
    <version_command><![CDATA[
        icescreen --version
        ]]>
    </version_command>
    <command detect_errors="aggressive"><![CDATA[
        #set $gbname = 'galaxy_genbank'
        mkdir -p ./tmp_icescreen/source_genbank
        && ln -s '${genome}' ./tmp_icescreen/source_genbank/${gbname}.gb
        && icescreen
            --galaxy
            --gbdir ./tmp_icescreen/source_genbank
            --outdir ./tmp_icescreen
            --phylum '${taxonomy.phylum}'
            --jobs "\${GALAXY_SLOTS:-4}" > ./tmp_icescreen/ICEscreen_log.txt 2>&1

        && cat ./tmp_icescreen/ICEscreen_results/results/${gbname}/icescreen_detection_ME/${gbname}_detected_ME.summary > '${summary}'
        && cat ./tmp_icescreen/ICEscreen_results/results/${gbname}/icescreen_detection_ME/${gbname}_detected_SP_withMEIds.tsv > '${detected_sp}'
        && cat ./tmp_icescreen/ICEscreen_results/results/${gbname}/icescreen_detection_ME/${gbname}_detected_ME.tsv > '${detected_me}'

        #if "output_annotation_genbank" in $additional_output.optional_files:
            && mv ./tmp_icescreen/ICEscreen_results/results/${gbname}/visualization_files/${gbname}_icescreen.gb '${gbout}'
        #end if
        #if "output_annotation_gff" in $additional_output.optional_files:
            && mv ./tmp_icescreen/ICEscreen_results/results/${gbname}/visualization_files/${gbname}_icescreen.gff '${gffout}'
        #end if
        #if "output_annotation_embl" in $additional_output.optional_files:
            && mv ./tmp_icescreen/ICEscreen_results/results/${gbname}/visualization_files/${gbname}_icescreen.embl '${emblout}'
        #end if
        #if "output_log" in $additional_output.optional_files:
            && mv ./tmp_icescreen/ICEscreen_log.txt '${log}'
        #end if
        #if "output_zip_all_files" in $additional_output.optional_files:
            && cd ./tmp_icescreen/ICEscreen_results/results/${gbname} && zip --quiet -r - -- . > '${outzip}'
        #end if
        ]]>
    </command>
    <inputs>
        <param name="genome" type="data" format="genbank" label="Input genomes to analyze in Genbank format" help="Multi-genbank files (i.e. gbff) are supported. Each record must include the ORIGIN nucleotide sequence at the end."/>
        <section name="taxonomy" title="Phylum of the genomes to analyse" expanded="True">
            <param name="phylum" type="select" label="Bacillota is the defaut parameter">
                <option value="bacillota" selected="true">Bacillota</option>
            </param>
        </section>
        <section name="additional_output" title="Additional output files" expanded="False">
            <param name="optional_files" type="select" label="Make the following optional files available in the Galaxy history" multiple="true" optional="true" display="checkboxes" >
                <option value="output_annotation_genbank">Genome annotation in genbank format</option>
                <option value="output_annotation_gff">Genome annotation in gff format</option>
                <option value="output_annotation_embl">Genome annotation in embl format</option>
                <option value="output_zip_all_files">Zip of all final and intermediate results</option>
                <option value="output_log">ICEscreen log</option>
            </param>
        </section>
    </inputs>
    <outputs>
      <data format="tabular" name="detected_sp" label="${tool.name} on ${on_string}: Signature Proteins table" />
      <data format="tabular" name="detected_me" label="${tool.name} on ${on_string}: ICEs/IMEs table" />
      <data format="txt" name="summary" label="${tool.name} on ${on_string}: results summary" />
      <data format="genbank" name="gbout" label="${tool.name} on ${on_string}: annotated genbank" >
          <filter>additional_output['optional_files'] and "output_annotation_genbank" in additional_output['optional_files']</filter>
      </data>
      <data format="gff3" name="gffout" label="${tool.name} on ${on_string}: annotated GFF3" >
          <filter>additional_output['optional_files'] and "output_annotation_gff" in additional_output['optional_files']</filter>
      </data>
      <data format="embl" name="emblout" label="${tool.name} on ${on_string}: annotated EMBL" >
          <filter>additional_output['optional_files'] and "output_annotation_embl" in additional_output['optional_files']</filter>
      </data>
      <data format="zip" name="outzip" label="${tool.name} on ${on_string}: all results zipped">
          <filter>additional_output['optional_files'] and "output_zip_all_files" in additional_output['optional_files']</filter>
      </data>
      <data format="txt" name="log" label="${tool.name} on ${on_string}: log file" >
          <filter>additional_output['optional_files'] and "output_log" in additional_output['optional_files']</filter>
      </data>
    </outputs>
    <tests>
        <test expect_num_outputs="3">
            <param name="genome" value="genbank/NC_004668_137848_164286.gb" ftype="genbank" />

            </output>
        </test>
        <test expect_num_outputs="4">
            <param name="genome" value="genbank/NC_004668_137848_164286.gb" ftype="genbank" />
            <param name="optional_files" value="output_annotation_genbank" />
            <output name="gbout" file="NC_004668_137848_164286_icescreen.gb" ftype="genbank" />
        </test>
        <test expect_num_outputs="4">
            <param name="genome" value="genbank/NC_004668_137848_164286.gb" ftype="genbank" />
            <param name="optional_files" value="output_annotation_gff" />
            <output name="gffout" file="NC_004668_137848_164286_icescreen.gff" ftype="gff3" />
        </test>
        <test expect_num_outputs="4">
            <param name="genome" value="genbank/NC_004668_137848_164286.gb" ftype="genbank" />
            <param name="optional_files" value="output_annotation_embl" />
            <output name="emblout" file="NC_004668_137848_164286_icescreen.embl" ftype="embl" />
        </test>
        <test expect_num_outputs="4">
            <param name="genome" value="genbank/NC_004668_137848_164286.gb" ftype="genbank" />
            <param name="optional_files" value="output_zip_all_files" />
            <output name="outzip" >
                <assert_contents>
                    <has_archive_member path=".*/*_detected_SP_withMEIds.tsv"><has_text text="ICE_IME_id" /></has_archive_member>
                </assert_contents>
            </output>
        </test>
        <test expect_num_outputs="4">
            <param name="genome" value="genbank/NC_004668_137848_164286.gb" ftype="genbank" />

                    <has_text text="bacillota" />
                    <not_has_text text="Error" />
                </assert_contents>
            </output>
        </test>
        <test expect_num_outputs="5">
            <param name="genome" value="genbank/NC_004668_137848_164286.gb" ftype="genbank" />
            <param name="phylum" value="bacillota" />
            <param name="optional_files" value="output_zip_all_files,output_log" />
            <output name="log" >
                <assert_contents>
                    <has_text text="bacillota" />
                    <not_has_text text="Error" />
                </assert_contents>

        </test>
    </tests>
    <help><![CDATA[
        .. class:: warningmark

        ICEscreen requires input files in genbank format. Multigenbank files (i.e. gbff files featuring multiple genome records back to back) are supported. Each Genbank record must include the ORIGIN nucleotide sequence.

-----

        **What it does**

        - Detected ICEs/IMEs table (`*_detected_ME.tsv`): list of the ICEs and IMEs elements detected by the tool, including information about the signature proteins they contain. Information in this file is complementary to the _withICEIMEIds.csv file above with each line representing an ICE / IME structures instead of a signature protein.
        - Results summary (`*_detected_ME.summary`): this file summarizes the main parameters and statistics regarding the ICE / IME structures and the SPs.

        Other optional and additional output files generated by ICEscreen:

        - Annotated Genbank, EMBL, and GFF3 files: tose files contains the annotations of the original Genbank file in addition to annotations added by ICEscreen such as ICEscreen signature proteins and mobile elements.
        - All results zipped file: You can download this file locally and unzip it to have the complete output folder generated by the tool, including intermediate processing files. See https://icescreen.migale.inrae.fr for details.
        - Log file (`*_detected_ME.log`): this file contains the detailed internal details used by ICEscreen to generate the results and logs of each step of the ICEscreen tool pipeline.

    ]]>

    </help>
    <citations>
        <citation type="bibtex">@UNPUBLISHED{Kim07aninterior-point,
            author = {Julie Lao and Thomas Lacroix and Gérard Guédon and Charles Coluzzi and Nathalie Leblond-Bourget and Hélène Chiapello},
            title = {"See our latest publication at https://icescreen.migale.inrae.fr."}
        }
        </citation>
    </citations>
</tool>

9:cca2a89210ae

<tool id="icescreen" name="ICEscreen" version="@TOOL_VERSION@+galaxy@VERSION_SUFFIX@" profile="20.09">
    <description>annotates ICEs and IMEs in Bacillota genomes</description>
    
    <macros>
        <import>macros.xml</import>
    </macros>
    <expand macro="xrefs"/>
    <expand macro="requirements"/>
    
    <version_command><![CDATA[
        icescreen --version
        ]]>
    </version_command>
    <command detect_errors="aggressive"><![CDATA[
        #set $gbname = 'galaxy_genbank'
        mkdir -p ./tmp_icescreen/source_genbank
        && mkdir -p ./tmp_icescreen/icescreen_result
        && ln -s '${genome}' ./tmp_icescreen/source_genbank/${gbname}.gb
        && icescreen
            --galaxy
            --gbdir ./tmp_icescreen/source_genbank
            --outdir ./tmp_icescreen/icescreen_result
            --phylum '${taxonomy.phylum}'
            --jobs "\${GALAXY_SLOTS:-4}" > ./tmp_icescreen/ICEscreen_log.txt 2>&1
            
        && cat ./tmp_icescreen/icescreen_result/ICEscreen_results/${gbname}/detected_mobile_elements/${gbname}_detected_ME.summary > '${summary}'
        && cat ./tmp_icescreen/icescreen_result/ICEscreen_results/${gbname}/detected_mobile_elements/${gbname}_detected_SP_withMEIds.tsv > '${detected_sp}'
        && cat ./tmp_icescreen/icescreen_result/ICEscreen_results/${gbname}/detected_mobile_elements/${gbname}_detected_ME.tsv > '${detected_me}'

        ]]>
    </command>
    <inputs>
        <param name="genome" type="data" format="genbank" label="Input genomes to analyze in Genbank format" help="Multi-genbank files (i.e. gbff) are supported. Each record must include the nucleotide sequence."/>
        <section name="taxonomy" title="Phylum of the genomes to analyse" expanded="True">
            <param name="phylum" type="select" label="Bacillota is the defaut parameter">
                <option value="bacillota" selected="true">Bacillota</option>
            </param>
        </section>
        <section name="additional_output" title="Additional output files" expanded="False">
            <param name="optional_files" type="select" label="Make the following optional files available in the Galaxy history" multiple="true" optional="true" display="checkboxes" >
                <option value="output_annotation_genbank" selected="false">Annotation in genbank format (full)</option>
                <option value="output_annotation_gff" selected="false">Annotation in gff format (ICEs and IMEs only)</option>
                <option value="output_annotation_embl" selected="false">Annotation in embl format (ICEs and IMEs only)</option>
                <option value="output_tar_gz_intermediate_files" selected="false">Archive of intermediate files</option>
                <option value="output_log" selected="false">Log file</option>
                <option value="output_param_conf" selected="false">Param conf file</option>
            </param>
        </section>
    </inputs>
    <outputs>
      <data format="tabular" name="detected_sp" label="${tool.name} on ${on_string}: Signature Proteins table" />
      <data format="tabular" name="detected_me" label="${tool.name} on ${on_string}: ICEs/IMEs table" />
      <data format="txt" name="summary" label="${tool.name} on ${on_string}: results summary" />
      <data format="genbank.gz" name="gbout" from_work_dir="./tmp_icescreen/icescreen_result/ICEscreen_results/galaxy_genbank/detected_mobile_elements/standard_genome_annotation_formats/galaxy_genbank_icescreen.gb.gz" label="${tool.name} on ${on_string}: genbank annotation (full)" >
          <filter>additional_output['optional_files'] and "output_annotation_genbank" in additional_output['optional_files']</filter>
      </data>
      <data format="gff3.gz" name="gffout" from_work_dir="./tmp_icescreen/icescreen_result/ICEscreen_results/galaxy_genbank/detected_mobile_elements/standard_genome_annotation_formats/galaxy_genbank_icescreen.gff.gz" label="${tool.name} on ${on_string}: GFF3 annotation (ICEs and IMEs only)" >
          <filter>additional_output['optional_files'] and "output_annotation_gff" in additional_output['optional_files']</filter>
      </data>
      <data format="embl.gz" name="emblout" from_work_dir="./tmp_icescreen/icescreen_result/ICEscreen_results/galaxy_genbank/detected_mobile_elements/standard_genome_annotation_formats/galaxy_genbank_icescreen.embl.gz" label="${tool.name} on ${on_string}: EMBL annotation (ICEs and IMEs only)" >
          <filter>additional_output['optional_files'] and "output_annotation_embl" in additional_output['optional_files']</filter>          
      </data>
      <data format="tar.gz" name="intermediate_tar_gz" from_work_dir="./tmp_icescreen/icescreen_result/ICEscreen_results/galaxy_genbank/tmp_intermediate_files.tar.gz" label="${tool.name} on ${on_string}: intermediate files archive">
          <filter>additional_output['optional_files'] and "output_tar_gz_intermediate_files" in additional_output['optional_files']</filter>
      </data>
      <data format="txt" name="log" from_work_dir="./tmp_icescreen/ICEscreen_log.txt" label="${tool.name} on ${on_string}: log file" >
          <filter>additional_output['optional_files'] and "output_log" in additional_output['optional_files']</filter>
      </data>
      <data format="gz" name="param_conf" from_work_dir="./tmp_icescreen/icescreen_result/ICEscreen_results/galaxy_genbank/param.conf.gz" label="${tool.name} on ${on_string}: param conf file" >
          <filter>additional_output['optional_files'] and "output_param_conf" in additional_output['optional_files']</filter>
      </data>
    </outputs>
    <tests>
        <test expect_num_outputs="3">
            <param name="genome" value="genbank/NC_004668_137848_164286.gb" ftype="genbank" />

            </output>
        </test>
        <test expect_num_outputs="4">
            <param name="genome" value="genbank/NC_004668_137848_164286.gb" ftype="genbank" />
            <param name="optional_files" value="output_annotation_genbank" />
            <output name="gbout" ftype="genbank.gz">
                <assert_contents>
                    <has_size value="19794" delta="100"/>
                </assert_contents>
            </output>
        </test>
        <test expect_num_outputs="4">
            <param name="genome" value="genbank/NC_004668_137848_164286.gb" ftype="genbank" />
            <param name="optional_files" value="output_annotation_gff" />
            <output name="gffout" ftype="gff3.gz">
                <assert_contents>
                    <has_size value="794" delta="100"/>
                </assert_contents>
            </output>
        </test>
        <test expect_num_outputs="4">
            <param name="genome" value="genbank/NC_004668_137848_164286.gb" ftype="genbank" />
            <param name="optional_files" value="output_annotation_embl" />
            <output name="emblout" ftype="embl.gz">
                <assert_contents>
                    <has_size value="1002" delta="100"/>
                </assert_contents>
            </output>
        </test>
        <test expect_num_outputs="4">
            <param name="genome" value="genbank/NC_004668_137848_164286.gb" ftype="genbank" />
            <param name="optional_files" value="output_tar_gz_intermediate_files" />
            <output name="intermediate_tar_gz" >
                <assert_contents>
                    <has_archive_member path=".*/*_detected_SP.tsv"><has_text text="Is_hit_blast" /></has_archive_member>
                </assert_contents>
            </output>
        </test>
        <test expect_num_outputs="4">
            <param name="genome" value="genbank/NC_004668_137848_164286.gb" ftype="genbank" />

                    <has_text text="bacillota" />
                    <not_has_text text="Error" />
                </assert_contents>
            </output>
        </test>
        <test expect_num_outputs="6">
            <param name="genome" value="genbank/NC_004668_137848_164286.gb" ftype="genbank" />
            <param name="phylum" value="bacillota" />
            <param name="optional_files" value="output_tar_gz_intermediate_files,output_log,output_param_conf" />
            <output name="log" >
                <assert_contents>
                    <has_text text="bacillota" />
                    <not_has_text text="Error" />
                </assert_contents>

        </test>
    </tests>
    <help><![CDATA[
        .. class:: warningmark

        ICEscreen requires input files in genbank format. Multigenbank files (i.e. gbff files featuring multiple genome records back to back) are supported. Each Genbank record must include the nucleotide sequence.

-----

        **What it does**

        - Detected ICEs/IMEs table (`*_detected_ME.tsv`): list of the ICEs and IMEs elements detected by the tool, including information about the signature proteins they contain. Information in this file is complementary to the _withICEIMEIds.csv file above with each line representing an ICE / IME structures instead of a signature protein.
        - Results summary (`*_detected_ME.summary`): this file summarizes the main parameters and statistics regarding the ICE / IME structures and the SPs.

        Other optional and additional output files generated by ICEscreen:

        - Genbank annotation file (full): annotations of the original Genbank file as well as annotations of ICEscreen on ICEs, IMEs, and the detected signature proteins.
        - EMBL and GFF3 annotation files (ICEs and IMEs only): annotations of ICEscreen on ICEs, IMEs, and the detected signature proteins.
        - Intermediate files (tar.gz): You can download this file locally and uncompress it to have the intermediate files generated by the tool. See https://icescreen.migale.inrae.fr for details.
        - Log file (`*_detected_ME.log`): this file contains the log of each step of the ICEscreen pipeline.
        - Param conf file (`param.conf.gz`): this file contains the configuration parameters related to the ICEscreen run.

    ]]>

    </help>

    <expand macro="citation"/>
    
</tool>