<tool id="domestication" name="Domestication Of New Parts" version="@TOOL_VERSION@+galaxy@VERSION_SUFFIX@" profile="21.09">
    <description>Adding the flanking enzyme sites to each DNA part</description>
    <macros>
        <token name="@VERSION_SUFFIX@">1</token>
        <token name="@TOOL_VERSION@">0.1.0</token>
    </macros>
    <requirements>
        <requirement type="package" version="0.1.11">flametree</requirement>
        <requirement type="package" version="1.85"> biopython </requirement>
        <requirement type="package" version="0.1.10">proglog</requirement>
        <requirement type="package" version="0.2.2">genedom</requirement>
        <requirement type="package" version="2.2.3">pandas</requirement>
        <requirement type="package" version="2.0.12">dnacauldron</requirement>
        <requirement type="package" version="0.3.9">pdf-reports</requirement>
        <requirement type="package" version="0.1.8">sequenticon</requirement>
        <requirement type="package" version="3.1.5">dna_features_viewer</requirement>
    </requirements>
    <command detect_errors="exit_code"><![CDATA[
        #set genbank_file_paths = ','.join([str(f) for f in $genbank_files])
        #set $file_name_mapping = ",".join(["%s:%s" % (file.file_name, file.name) for file in $genbank_files])
        mkdir 'metoutdir_gb' && mkdir 'outdir_zip' && mkdir 'outdir_gb'&&
        cp '$domestication_csv' 'domestication_csv.csv' &&
        python '$__tool_directory__/domestication.py'
            --files_to_domestication '$genbank_file_paths'
            --csv_file 'domestication_csv.csv'
            --file_name_mapping '$file_name_mapping'
            --use_file_names_as_id '$adv.use_file_names_as_ids'
            --allow_edits '$adv.allow_edits'
            --output_dom 'output_zip.zip'
            --output_gb_dom 'outdir_gb'
            --output_methprot 'metoutdir_gb'
            --methylation_protection '$methylation_protection' &&
        cp 'output_zip.zip' '$domesticated_zip'
    ]]></command>
    <inputs>
        <param name="genbank_files" type="data_collection" collection_type="list" format="genbank" label="GenBank File(s)"/>
        <param name="domestication_csv" type="data" format="csv" label="Deomestication csv"/>
        <param name="methylation_protection" type="boolean" checked="False" label="Enabel Methylation Protection"/>
        <section name="adv" title="Advanced Options" expanded="false">
            <param name="use_file_names_as_ids" type="boolean" checked="True" label="Use File Names As Sequence IDs" />
            <param name="allow_edits" type="boolean" checked="True" label="Allow Sequence Edits" help="If False, sequences cannot be edited by the domesticator, only extended
      with flanks. If a sequence has for instance forbidden restriction sites,
      the domesticaton will fail for this sequence (and this will be noted in
      the report."/>
        </section>
    </inputs>   
    <outputs>
        <data format="zip" name="domesticated_zip" label="domestication results"/>
        <collection name="domesticated_gb" type="list" label="Domesticated GenBank Files">
            <filter>methylation_protection == False</filter>
            <discover_datasets pattern="(?P&lt;name&gt;.*).gb" format="genbank" directory="outdir_gb" />
        </collection>
        <collection name="methprot_gb" type="list" label="GenBank Files methylation protected">
            <filter>methylation_protection == True</filter>
            <discover_datasets pattern="(?P&lt;name&gt;.*).gb" format="genbank" directory="metoutdir_gb" />
        </collection>
    </outputs>
    <tests>
        <!--test without methylation protection-->
        <test expect_num_outputs="2"> 
            <param name="genbank_files">
                <collection type="list">
                    <element name="p3_2_CAG" value="EMMA_undomesticated_parts/p3_2_CAG.gb" />
                    <element name="p7_gfp_sequence" value="EMMA_undomesticated_parts/p7_gfp_sequence.gb" />
                    <element name="p14_CMVp" value="EMMA_undomesticated_parts/p14_CMVp.gb" />
                    <element name="p16_bGHpolyA" value="EMMA_undomesticated_parts/p16_bGHpolyA.gb" />
                    <element name="p18_CMVp" value="EMMA_undomesticated_parts/p18_CMVp.gb" />
                    <element name="p19_mNeoGreen" value="EMMA_undomesticated_parts/p19_mNeoGreen.gb" />
                    <element name="p22_PGKpolyA" value="EMMA_undomesticated_parts/p22_PGKpolyA.gb" />
                </collection>
            </param>
            <param name="domestication_csv" value="EMMA.csv" />
            <param name='methylation_protection' value='false' />
            <param name="adv|use_file_names_as_ids" value="True" />
            <param name="adv|allow_edits" value="True" />
            <output name="domesticated_zip" ftype='zip'>
                <assert_contents>
                    <has_archive_member path=".*" n="24"/>
                    <has_archive_member path="order_ids.csv">
                        <has_n_lines n="8" />
                    </has_archive_member>
                </assert_contents>
            </output>
            <output_collection name="domesticated_gb" type="list" count="7">
                <element name="p14_CMVp">
                    <assert_contents>
                     <has_n_lines min="20" />
                    </assert_contents>
                </element>
                <element name="p16_bGHpolyA">
                    <assert_contents>
                     <has_n_lines min="20" />
                    </assert_contents>
                </element>
                <element name="p18_CMVp">
                    <assert_contents>
                     <has_n_lines min="20" />
                    </assert_contents>
                </element>
                 <element name="p19_mNeoGreen">
                    <assert_contents>
                     <has_n_lines min="20" />
                    </assert_contents>
                </element>
                <element name="p22_PGKpolyA">
                    <assert_contents>
                     <has_n_lines min="20" />
                    </assert_contents>
                </element>
                <element name="p3_2_CAG">
                    <assert_contents>
                     <has_n_lines min="20" />
                    </assert_contents>
                </element>
                <element name="p7_gfp_sequence">
                    <assert_contents>
                     <has_n_lines min="20" />
                    </assert_contents>
                </element>
            </output_collection>
        </test>
        <!--test with methylation protection-->
        <test expect_num_outputs="2"> 
            <param name="genbank_files">
                <collection type="list">
                    <element name="p3_2_CAG" value="EMMA_undomesticated_parts/p3_2_CAG.gb" />
                    <element name="p7_gfp_sequence" value="EMMA_undomesticated_parts/p7_gfp_sequence.gb" />
                    <element name="p14_CMVp" value="EMMA_undomesticated_parts/p14_CMVp.gb" />
                    <element name="p16_bGHpolyA" value="EMMA_undomesticated_parts/p16_bGHpolyA.gb" />
                    <element name="p18_CMVp" value="EMMA_undomesticated_parts/p18_CMVp.gb" />
                    <element name="p19_mNeoGreen" value="EMMA_undomesticated_parts/p19_mNeoGreen.gb" />
                    <element name="p22_PGKpolyA" value="EMMA_undomesticated_parts/p22_PGKpolyA.gb" />
                </collection>
            </param>
            <param name="domestication_csv" value="EMMA.csv" />
            <param name="methylation_protection" value="true" />
            <param name="adv|use_file_names_as_ids" value="True" />
            <param name="adv|allow_edits" value="True" />
            <output name="domesticated_zip" ftype='zip'>
                <assert_contents>
                    <has_archive_member path=".*" n="24"/>
                    <has_archive_member path="order_ids.csv">
                        <has_n_lines n="8" />
                    </has_archive_member>
                </assert_contents>
            </output>
            <output_collection name="methprot_gb" type="list" count="7">
                <element name="p14_CMVp">
                    <assert_contents>
                     <has_n_lines min="20" />
                    </assert_contents>
                </element>
                <element name="p16_bGHpolyA">
                    <assert_contents>
                     <has_n_lines min="20" />
                    </assert_contents>
                </element>
                <element name="p18_CMVp">
                    <assert_contents>
                     <has_n_lines min="20" />
                    </assert_contents>
                </element>
                 <element name="p19_mNeoGreen">
                    <assert_contents>
                     <has_n_lines min="20" />
                    </assert_contents>
                </element>
                <element name="p22_PGKpolyA">
                    <assert_contents>
                     <has_n_lines min="20" />
                    </assert_contents>
                </element>
                <element name="p3_2_CAG">
                    <assert_contents>
                     <has_n_lines min="20" />
                    </assert_contents>
                </element>
                <element name="p7_gfp_sequence">
                    <assert_contents>
                     <has_n_lines min="20" />
                    </assert_contents>
                </element>
            </output_collection>
        </test>
    </tests>
    <help><![CDATA[
Domestication Of New Parts
==========================

The Domestication tool from the EGF Biofoundry, powered by Genedom, enables rapid and accurate add of the flanking enzyme sites to each DNA part preparing it for modular cloning systems, such as Golden Gate Assembly (complete documentation `here <https://edinburgh-genome-foundry.github.io/genedom/>`_)

**Parameters**:
---------------
* **GenBank File(s)**: GenBank files for domestication (can be output of sculpt tool)
* **Deomestication csv**: csv file containes the construct names with their parts.`example <https://cuba.genomefoundry.org/static/file_examples/domesticate_part_batches/EMMA.csv>`_
* **Enabel Methylation Protection**: Enable the methylation protection.
* **Allow Sequence Edits**: Recommanded. If False, sequences cannot be edited by the domesticator, only extended with flanks. If a sequence has for instance forbidden restriction sites, the domesticaton will fail for this sequence (and this will be noted in the report.
* **Use File Names As Sequence IDs**: Recommended if the GenBank file names represent the fragment names.
* **output**: Zip report for domestication process + domesticated gb files (methylation pretected in case of enable of methylation protecton option).
    ]]></help>
    <citations>
        <citation type="bibtex">
            @unpublished{domestication
                author = {Ramiz Khaled},
                title = {{domestication}},
                url = {https://github.com/Edinburgh-Genome-Foundry/Examples/blob/master/templates/template1.ipynb},
            }
        </citation>
    </citations>
</tool>