<?xml version="1.0"?>
<tool id="dada2_fetcher" name="dada2 data manager" tool_type="manage_data" version="0.1.0">
    <description>Download reference databases</description>
    <requirements>
        <requirement type="package" version="3.7">python</requirement>
    </requirements>
    <command detect_errors="exit_code"><![CDATA[
    python '$__tool_directory__/data_manager.py'
    --out '$out_file'
    #set dataset = str($db_cond.db_select) + '_' + str($db_cond.version_select)
    --dataset '$dataset'
    ]]>
    </command>
    <inputs>
        <conditional name="db_cond">
            <param name="db_select" type="select" label="Taxonomic database">
                <option value="silva">Silva</option>
                <option value="rdp">RDP</option>
                <option value="greengenes">GreenGenes</option>
                <option value="unite">UNITE Fungi: General Fasta</option>
                <!-- UNITE Eukaryotes not yet supported https://github.com/benjjneb/dada2/issues/702 -->
                <option value="RefSeq_RDP">NCBI RefSeq 16S rRNA database supplemented by RDP</option>
                <option value="gtdb">GTDB: Genome Taxonomy Database (Bacteria &amp; Archaea)</option>
                <option value="hitdb">HitDB (Human InTestinal 16S)</option>
                <option value="silva_euk_18S">Silva Eukaryotic 18S</option>
                <option value="PR2">Protist Ribosomal Reference database (PR2)</option>
            </param>
            <when value="silva">
                <param name="version_select" type="select" label="Database version">
                    <option value="138">138</option>
                    <option value="132">132</option>
                    <option value="128">128</option>
                </param>
            </when>
            <when value="rdp">
                <param name="version_select" type="select" label="Database version">
                    <option value="16">16</option>
                    <option value="14">14</option>
                </param>
            </when>
            <when value="greengenes">
                <param name="version_select" type="select" label="Database version">
                    <option value="13.84">13.84</option>
                </param>
            </when>
            <when value="unite">
                <param name="version_select" type="select" label="Database version">
                    <option value="8.0_fungi">release 8.0 for Fungi</option>
                    <option value="8.0_fungi_singletons">release 8.0 for Fungi including global and 97% singletons</option>
                </param>
            </when>
            <when value="RefSeq_RDP">
                <param name="version_select" type="select" label="Database version">
                    <option value="2018_05">05/2018</option>
                </param>
            </when>
            <when value="gtdb">
                <param name="version_select" type="select" label="Database version">
                    <option value="2018_11">11/2018</option>
                </param>
            </when>
            <when value="hitdb">
                <param name="version_select" type="select" label="Database version">
                    <option value="1">1</option>
                </param>
            </when>
            <when value="silva_euk_18S">
                <param name="version_select" type="select" label="Database version">
                    <option value="132">132</option>
                </param>
            </when>
            <when value="PR2">
                <param name="version_select" type="select" label="Database version">
                    <option value="4.11.1">4.11.1</option>
                </param>
            </when>
        </conditional>
    </inputs>
    <outputs>
        <data name="out_file" format="data_manager_json" />
    </outputs>
    <tests>
        <test>
            <param name="db_cond|db_select" value="silva"/>
            <param name="db_cond|version_select" value="138"/>
            <output name="out_file" file="silva138_json"/>
        </test>
        <test>
            <param name="db_cond|db_select" value="silva"/>
            <param name="db_cond|version_select" value="132"/>
            <output name="out_file" file="silva132_json"/>
        </test>
        <test>
            <param name="db_cond|db_select" value="rdp"/>
            <param name="db_cond|version_select" value="16"/>
            <output name="out_file" file="rdp16_json"/>
        </test>
        <test>
            <param name="db_cond|db_select" value="greengenes"/>
            <param name="db_cond|version_select" value="13.84"/>
            <output name="out_file" file="greengenes13.84_json"/>
        </test>
        <test>
            <param name="db_cond|db_select" value="unite"/>
            <param name="db_cond|version_select" value="8.0_fungi"/>
            <output name="out_file" file="unite8fungi_json"/>
        </test>
        <test>
            <param name="db_cond|db_select" value="unite"/>
            <param name="db_cond|version_select" value="8.0_fungi_singletons"/>
            <output name="out_file" file="unite8fungisingletons_json"/>
        </test>
        <test>
            <param name="db_cond|db_select" value="RefSeq_RDP"/>
            <param name="db_cond|version_select" value="2018_05"/>
            <output name="out_file" file="RefSeq_RDP2018_json"/>
        </test>
        <test>
            <param name="db_cond|db_select" value="gtdb"/>
            <param name="db_cond|version_select" value="2018_11"/>
            <output name="out_file" file="gtdb2018_json"/>
        </test>
        <test>
            <param name="db_cond|db_select" value="hitdb"/>
            <param name="db_cond|version_select" value="1"/>
            <output name="out_file" file="hitdb1_json"/>
        </test>
        <test>
            <param name="db_cond|db_select" value="silva_euk_18S"/>
            <param name="db_cond|version_select" value="132"/>
            <output name="out_file" file="silvaeuk132_json"/>
        </test>
        <test>
            <param name="db_cond|db_select" value="PR2"/>
            <param name="db_cond|version_select" value="4.11.1"/>
            <output name="out_file" file="PR24.11.1_json"/>
        </test>
    </tests>
    <help><![CDATA[
Public Reference databases maintained by the DADA2 project
..........................................................

The following refrence databases which are describes as maintained by the DADA2 project (https://benjjneb.github.io/dada2/training.html) are available

- Silva (https://www.arb-silva.de/)
- RDP
- GreenGenes (http://greengenes.secondgenome.com/)
- UNITE general FASTA (https://unite.ut.ee/repository.php)

While Silva and RDP contain reference databases for taxonomy and species assignment, the greengenes and UNITE databases only contains a reference database for taxonomy assignment.

For the Silva databases check the license information: http://www.arb-silva.de/silva-license-information.

Except for UNITE all reference databases are downloaded from the corresponding zenodo links that are listed on the DADA2 website. The UNITE databases are taken from the links provided on the UNITE website

More detailed informations in the reference data bases can be found on the DADA2 website and contained links: https://benjjneb.github.io/dada2/training.html.

Further public Reference databases listed by the DADA2 project
..............................................................

Several contributed reference databases are listed of the DADA2 project website (https://benjjneb.github.io/dada2/training.html):

- RefSeq + RDP (NCBI RefSeq 16S rRNA database supplemented by RDP)
- GTDB: Genome Taxonomy Database (More info: http://gtdb.ecogenomic.org/)
- HitDB version 1 (Human InTestinal 16S rRNA) (https://github.com/microbiome/HITdb)
- RDP fungi LSU
- Silva Eukaryotic 18S
- PR2 (https://github.com/pr2database/pr2database)

Except for PR2, all reference databases are downloaded from the corresponding zenodo links that are listed on the DADA2 website. The PR2 database is taken from their github page.

More detailed informations in the reference data bases can be found on the DADA2 website and contained links: https://benjjneb.github.io/dada2/training.html.
    ]]></help>
    <citations>
        <!-- silva -->
         <citation type="doi">10.1093/nar/gks1219</citation>
        <!-- rdp -->
        <citation type="doi">10.1093/nar/gkt1244</citation>
        <!-- greengenes -->
        <citation type="doi">10.1128/AEM.03006-05</citation>
        <!-- unite -->
        <citation type="doi">10.15156/BIO/786343</citation>
        <!-- TODO gtdb ??? -->
        <!-- hitdb -->
        <citation type="doi">10.1186/s12864-015-2265-y</citation>
        <!-- PR2 -->
        <citation type="doi">10.1093/nar/gks1160</citation>
    </citations>
</tool>