annotate data_manager/kraken2_build_database.xml @ 14:8c533e19b697 draft default tip

planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_build_kraken2_database/ commit 66be7bc4e07926c3645ba87f78702f2267fad7b3

<tool id="kraken2_build_database" name="Kraken2" tool_type="manage_data" version="@TOOL_VERSION@+galaxy@VERSION_SUFFIX@" profile="@PROFILE@">
    <description>database builder</description>
    <macros>
        <token name="@TOOL_VERSION@">2.1.3</token>
        <token name="@VERSION_SUFFIX@">5</token>
        <token name="@PROFILE@">22.01</token>
        <xml name="common_params">
            <param name="kmer_len" type="integer" value="35" label="K-mer length in BP" />
            <param name="minimizer_len" type="integer" value="31" label="Minimizer length" />
            <param name="minimizer_spaces" type="integer" value="7" label="Minimizer spaces" />
            <param name="load_factor" type="float" value="0.7" min="0" max="1" label="Load factor" help="Proportion of the hash table to be populated" />
            <param name="clean" type="boolean" truevalue="--clean" falsevalue="" checked="true" label="Clean up extra files" />
        </xml>
        <xml name="viral">
            <option value="viral">Viral (viral; ~0.5 GB)</option>
        </xml>
        <xml name="minusb">
            <option value="minusb">MinusB (archaea, viral, plasmid, human, UniVec_Core; ~8 GB)</option>
        </xml>
        <xml name="standard">
            <option value="standard">Standard-Full (archaea, bacteria, viral, plasmid, human, UniVec_Core; ~50 GB)</option>
        </xml>
        <xml name="standard_08gb">
            <option value="standard_08gb">Standard-8 (Standard with DB capped at 8 GB)</option>
        </xml>
        <xml name="standard_16gb">
            <option value="standard_16gb">Standard-16 (Standard with DB capped at 16 GB)</option>
        </xml>
        <xml name="pluspf">
            <option value="pluspf">PlusPF (Standard plus protozoa and fungi; ~50 GB)</option>
        </xml>
        <xml name="pluspf_08gb">
            <option value="pluspf_08gb">PlusPF-8 (PlusPF with DB capped at 8 GB; ~7.5 GB)</option>
        </xml>
        <xml name="pluspf_16gb">
            <option value="pluspf_16gb">PlusPF-16 (PlusPF with DB capped at 16 GB; ~15 GB)</option>
        </xml>
        <xml name="pluspfp">
            <option value="pluspfp">PlusPFP (Standard plus protozoa, fungi and plant; ~129 GB)</option>
        </xml>
        <xml name="pluspfp_08gb">
            <option value="pluspfp_08gb">PlusPFP-8 (PlusPFP with DB capped at 8 GB; ~7.5 GB)</option>
        </xml>
        <xml name="pluspfp_16gb">
            <option value="pluspfp_16gb">PlusPFP-16 (PlusPFP with DB capped at 16 GB; ~15 GB)</option>
        </xml>
        <xml name="core_nt">
            <option value="core_nt">core_nt (Very large collection, inclusive of GenBank, RefSeq, TPA and PDB; ~182 GB)</option>
        </xml>
    </macros>
    <xrefs>
        <xref type="bio.tools">kraken2</xref>
    </xrefs>
    <requirements>
        <requirement type="package" version="@TOOL_VERSION@">kraken2</requirement>
        <requirement type="package" version="3.13">python</requirement>
    </requirements>
    <version_command>kraken2 -version | head -n 1 | awk '{print $NF}'</version_command>
    <command detect_errors="exit_code"><![CDATA[
python '$__tool_directory__/kraken2_build_database.py'
    '$out_file'
    --database-type '$database_type.database_type'
#if $database_type.database_type == "standard_local_build"
    --threads \${GALAXY_SLOTS:-1}
    --kmer-len $database_type.kmer_len
    --minimizer-len $database_type.minimizer_len
    --minimizer-spaces $database_type.minimizer_spaces
    --load-factor $database_type.load_factor
    $database_type.clean
#else if $database_type.database_type == "standard_prebuilt"
    --prebuilt-db '$database_type.prebuild.prebuilt_db'
    --prebuilt-date '$database_type.prebuild.prebuilt_date'
#else if $database_type.database_type == "minikraken"
    --minikraken2-version '$database_type.minikraken2_version'
#else if $database_type.database_type == "special_prebuilt"
    --prebuilt-db '$database_type.special_prebuild.prebuilt_db'
    --prebuilt-date '$database_type.special_prebuild.prebuilt_date'
#else if $database_type.database_type == "special"
    --threads \${GALAXY_SLOTS:-1}
    --special-database-type '$database_type.special_database_type'
    --kmer-len $database_type.kmer_len
    --minimizer-len $database_type.minimizer_len
    --minimizer-spaces $database_type.minimizer_spaces
    --load-factor $database_type.load_factor
    $database_type.clean
#else if $database_type.database_type == "custom"
    --threads \${GALAXY_SLOTS:-1}
    --custom-fasta '$database_type.custom_fasta'
    --custom-database-name '$database_type.custom_database_name'
    --custom-source-info '$database_type.custom_source_info'
    $database_type.skip_maps
    --kmer-len $database_type.kmer_len
    --minimizer-len $database_type.minimizer_len
    --minimizer-spaces $database_type.minimizer_spaces
    --load-factor $database_type.load_factor
    $database_type.clean
#end if
]]>
    </command>
    <inputs>
        <conditional name="database_type">
            <param name="database_type" type="select" multiple="false" label="Database Type">
                <option value="standard_local_build">Standard, Local Build</option>
                <option value="standard_prebuilt">Pre-Built Refseq indexes</option>
                <option value="minikraken">MiniKraken</option>
                <option value="special_prebuilt">Special Pre-Built indexes</option>
                <option value="special">Special</option>
                <option value="custom">Custom</option>
            </param>
            <when value="standard_local_build">
                <expand macro="common_params" />
            </when>
            <when value="standard_prebuilt">
                <conditional name="prebuild">
                    <param name="prebuilt_date" type="select" label="Select index build date">
                        <option value="2024-09-04">September 4, 2024</option>
                        <option value="2024-06-05">June 5, 2024</option>
                        <option value="2024-01-12">January 12, 2024</option>
                        <option value="2023-06-05">June 5, 2023</option>
                        <option value="2022-06-07">June 7, 2022</option>
                        <option value="2021-05-17">May 17, 2021</option>
                        <option value="2021-01-27">January 27, 2021</option>
                        <option value="2020-12-02">December 2, 2020</option>
                        <option value="2020-09-19">September 19, 2020</option>
                    </param>
                    <when value="2024-09-04">
                        <param name="prebuilt_db" type="select" label="Select a prebuilt Refseq index to download">
                            <expand macro="viral"/>
                            <expand macro="minusb"/>
                            <expand macro="standard"/>
                            <expand macro="standard_08gb"/>
                            <expand macro="standard_16gb"/>
                            <expand macro="pluspf"/>
                            <expand macro="pluspf_08gb"/>
                            <expand macro="pluspf_16gb"/>
                            <expand macro="pluspfp"/>
                            <expand macro="pluspfp_08gb"/>
                            <expand macro="core_nt"/>
                        </param>
                    </when>
                    <when value="2024-06-05">
                        <param name="prebuilt_db" type="select" label="Select a prebuilt Refseq index to download">
                            <expand macro="viral"/>
                            <expand macro="minusb"/>
                            <expand macro="standard"/>
                            <expand macro="standard_08gb"/>
                            <expand macro="standard_16gb"/>
                            <expand macro="pluspf"/>
                            <expand macro="pluspf_08gb"/>
                            <expand macro="pluspf_16gb"/>
                            <expand macro="pluspfp"/>
                            <expand macro="pluspfp_08gb"/>
                            <expand macro="pluspfp_16gb"/>
                        </param>
                    </when>
                    <when value="2024-01-12">
                        <param name="prebuilt_db" type="select" label="Select a prebuilt Refseq index to download">
                            <expand macro="viral"/>
                            <expand macro="minusb"/>
                            <expand macro="standard"/>
                            <expand macro="standard_08gb"/>
                            <expand macro="standard_16gb"/>
                            <expand macro="pluspf"/>
                            <expand macro="pluspf_08gb"/>
                            <expand macro="pluspf_16gb"/>
                            <expand macro="pluspfp"/>
                            <expand macro="pluspfp_08gb"/>
                            <expand macro="pluspfp_16gb"/>
                        </param>
                    </when>
                    <when value="2023-06-05">
                        <param name="prebuilt_db" type="select" label="Select a prebuilt Refseq index to download">
                            <expand macro="viral"/>
                            <expand macro="minusb"/>
                            <expand macro="standard"/>
                            <expand macro="standard_08gb"/>
                            <expand macro="standard_16gb"/>
                            <expand macro="pluspf"/>
                            <expand macro="pluspf_08gb"/>
                            <expand macro="pluspf_16gb"/>
                            <expand macro="pluspfp"/>
                            <expand macro="pluspfp_08gb"/>
                            <expand macro="pluspfp_16gb"/>
                        </param>
                    </when>
                    <when value="2022-06-07">
                        <param name="prebuilt_db" type="select" label="Select a prebuilt Refseq index to download">
                            <expand macro="viral"/>
                            <expand macro="minusb"/>
                            <expand macro="standard"/>
                            <expand macro="standard_08gb"/>
                            <expand macro="standard_16gb"/>
                            <expand macro="pluspf"/>
                            <expand macro="pluspf_08gb"/>
                            <expand macro="pluspf_16gb"/>
                            <expand macro="pluspfp"/>
                            <expand macro="pluspfp_08gb"/>
                            <expand macro="pluspfp_16gb"/>
                        </param>
                    </when>
                    <when value="2021-05-17">
                        <param name="prebuilt_db" type="select" label="Select a prebuilt Refseq index to download">
                            <expand macro="viral"/>
                            <expand macro="minusb"/>
                            <expand macro="standard"/>
                            <expand macro="standard_08gb"/>
                            <expand macro="standard_16gb"/>
                            <expand macro="pluspf"/>
                            <expand macro="pluspf_08gb"/>
                            <expand macro="pluspf_16gb"/>
                            <expand macro="pluspfp_08gb"/>
                            <expand macro="pluspfp_16gb"/>
                        </param>
                    </when>
                    <when value="2021-01-27">
                        <param name="prebuilt_db" type="select" label="Select a prebuilt Refseq index to download">
                            <expand macro="pluspf"/>
                            <expand macro="pluspf_08gb"/>
                            <expand macro="pluspf_16gb"/>
                            <expand macro="pluspfp_08gb"/>
                            <expand macro="pluspfp_16gb"/>
                        </param>
                    </when>
                    <when value="2020-12-02">
                        <param name="prebuilt_db" type="select" label="Select a prebuilt Refseq index to download">
                            <expand macro="viral"/>
                            <expand macro="minusb"/>
                            <expand macro="standard"/>
                            <expand macro="standard_08gb"/>
                            <expand macro="standard_16gb"/>
                        </param>
                    </when>
                    <when value="2020-09-19">
                        <param name="prebuilt_db" type="select" multiple="false" label="Select a prebuilt Refseq indexes to download">
                            <expand macro="minusb"/>
                            <expand macro="standard"/>
                            <expand macro="standard_08gb"/>
                            <expand macro="standard_16gb"/>
                            <expand macro="pluspf"/>
                            <expand macro="pluspf_08gb"/>
                            <expand macro="pluspf_16gb"/>
                            <expand macro="pluspfp_08gb"/>
                            <expand macro="pluspfp_16gb"/>
                        </param>
                    </when>
                </conditional>
            </when>
            <when value="special_prebuilt">
                <conditional name="special_prebuild">
                    <param name="special_prebuilt_db" type="select" multiple="false" label="Select pre-built database to download">
                        <option value="gtdb_genome_reps_20241109">GTDB v220 (Bacterial and archaeal; ~497 GB) (December 13, 2024)</option>
                        <option value="eupathdb48_20230407">EuPathDB-46 (April 18, 2023)</option>
                        <option value="eupathdb48_20201113">EuPathDB-46 (November 13, 2020)</option>
                    </param>
                    <when value="gtdb_genome_reps_20241109">
                        <param name="prebuilt_db" type="hidden" value="gtdb_genome_reps"/>
                        <param name="prebuilt_date" type="hidden" value="2024-11-09"/>
                    </when>
                    <when value="eupathdb48_20230407">
                        <param name="prebuilt_db" type="hidden" value="eupathdb48"/>
                        <param name="prebuilt_date" type="hidden" value="2023-04-07"/>
                    </when>
                    <when value="eupathdb48_20201113">
                        <param name="prebuilt_db" type="hidden" value="eupathdb48"/>
                        <param name="prebuilt_date" type="hidden" value="2020-11-13"/>
                    </when>
                </conditional>
            </when>
            <when value="minikraken">
                <param name="minikraken2_version" type="select" multiple="false" label="Select MiniKraken2 database version to download">
                    <option value="v2">Version 2</option>
                    <option value="v1">Version 1</option>
                </param>
            </when>
            <when value="special">
                <param name="special_database_type" type="select" multiple="false" label="Select database to build">
                    <option value="greengenes">Greengenes</option>
                    <option value="silva">Silva</option>
                    <option value="rdp">RDP</option>
                </param>
                <expand macro="common_params" />
            </when>
            <when value="custom">
                <param name="custom_fasta" type="data" format="fasta" multiple="False" label="Select history item" />
                <param name="custom_database_name" type="text" label="Name for this database" />
                <param name="custom_source_info" type="text" label="Database source info" help="Concise description of how this build has been sourced. This description will be appended (in parentheses) to the user-facing name of the build. Example: https://doi.org/10.5281/zenodo.8339822, from v1 assembly_summary.txt sequences" />
                <param name="skip_maps" type="boolean" truevalue="--skip-maps" falsevalue="" label="Skip downloading accession number to taxid maps during taxonomy download." />
                <expand macro="common_params" />
            </when>
        </conditional>
    </inputs>
    <outputs>
        <data name="out_file" format="data_manager_json" />
    </outputs>
    <tests>
        <test expect_num_outputs="1">
            <conditional name="database_type">
                <param name="database_type" value="custom" />
                <param name="custom_fasta" value="adapter.fa" />
                <param name="custom_database_name" value="custom_database" />
                <param name="custom_source_info" value="from adapter.fa test data" />
                <param name="skip_maps" value="true" />
                <param name="kmer_len" value="35" />
                <param name="minimizer_spaces" value="6"/>
                <param name="load_factor" value="0.7" />
                <param name="clean" value="true"/>
            </conditional>
            <output name="out_file">
                <assert_contents>
                    <has_text text="kraken2_databases"/>
                    <has_text text="path"/>
                    <has_text text="custom_database (from adapter.fa test data, kmer-len=35, minimizer-len=31, minimizer-spaces=6, load-factor=0.7)"/>
                </assert_contents>
            </output>
        </test>
        <test>
            <conditional name="database_type">
                <param name="database_type" value="standard_prebuilt" />
                <conditional name="prebuild">
                    <param name="prebuilt_date" value="2022-06-07"/>
                    <param name="prebuilt_db" value="viral"/>
                </conditional>
            </conditional>
            <output name="out_file">
                <assert_contents>
                    <has_text text="kraken2_databases"/>
                    <has_text text="path"/>
                    <has_text text="Viral"/>
                    <has_text text="_standard_prebuilt_viral_2022-06-07"/>
                    <has_text text="Version"/>
                    <has_text text="Downloaded"/>
                </assert_contents>
            </output>
        </test>
        <!-- test with latest version -->
        <test>
            <conditional name="database_type">
                <param name="database_type" value="standard_prebuilt" />
                <conditional name="prebuild">
                    <param name="prebuilt_date" value="2024-01-12"/>
                    <param name="prebuilt_db" value="viral"/>
                </conditional>
            </conditional>
            <output name="out_file">
                <assert_contents>
                    <has_text text="kraken2_databases"/>
                    <has_text text="path"/>
                    <has_text text="Viral"/>
                    <has_text text="_standard_prebuilt_viral_2024-01-12"/>
                    <has_text text="Version"/>
                    <has_text text="Downloaded"/>
                </assert_contents>
            </output>
        </test>
        <!-- test with first 2024 version -->
        <test>
            <conditional name="database_type">
                <param name="database_type" value="standard_prebuilt" />
                <conditional name="prebuild">
                    <param name="prebuilt_date" value="2024-06-05"/>
                    <param name="prebuilt_db" value="viral"/>
                </conditional>
            </conditional>
            <output name="out_file">
                <assert_contents>
                    <has_text text="kraken2_databases"/>
                    <has_text text="path"/>
                    <has_text text="Viral"/>
                    <has_text text="_standard_prebuilt_viral_2024-06-05"/>
                    <has_text text="Version"/>
                    <has_text text="Downloaded"/>
                </assert_contents>
            </output>
        </test>
    </tests>
    <help><![CDATA[
Build Kraken2 databases or download `prebuilt Kraken2 RefSeq indexes <https://benlangmead.github.io/aws-indexes/k2>`__
    ]]></help>
    <citations>
        <citation type="doi">10.1186/s13059-019-1891-0</citation>
    </citations>
</tool>