comparison pangolin.xml @ 28:9d61d3fa046c draft

planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/pangolin commit f93f921e3c2d0002ff0c152d90b9221533ad22e9

27:03ed9524dcd1

<tool id="pangolin" name="Pangolin" version="@TOOL_VERSION@+galaxy0" profile="20.01">
    <description>Phylogenetic Assignment of Outbreak Lineages</description>
    <macros>
        <token name="@TOOL_VERSION@">4.2</token>
        <token name="@PANGOLIN_DATA_VERSION@">1.17</token>
        <token name="@CONSTELLATIONS_VERSION@">0.1.10</token>
        <token name="@MIN_COMPATIBLE_PANGOLIN_DATA_FORMAT@">4</token>
        <!-- a regex describing the scorpio versions that this wrapper version
        is backwards-compatible with; can be used with the min_scorpio_version
        column of the constellations data table to offer only compatible
        versions of constellations data. -->
        <token name="@COMPATIBLE_SCORPIO_DATA_FORMAT@"><![CDATA[(^0\.[1-3]$|^0\.[0-2]\.\d+$|^0\.3\.\d$|^0\.3\.1[0-7]$|^0$)]]></token>
        <xml name="usher_download_option">
            <when value="download">
                <param argument="--use-assignment-cache" type="boolean" truevalue="--use-assignment-cache" falsevalue="" label="Download and use also latest UShER assignment cache?"
                help="Get the latest UShER assignment cache from the pangolin-assignment online repository and use it to speed up UShER lineage assignment. Note: Downloading the cached assignments will only pay off for large numbers of input samples." />
            </when>
        </xml>
        <xml name="cached_usher_assignment_cache">
            <param name="assignment_cache_release" type="select" optional="true" label="Use corresponding UShER assignment cache?"
            help="If the server offers a copy of the UShER assignment cache along with the specified version of pangolin-data, you can select it here to speed up UShER lineage assignment. If no suitable assignment cache is available, it is perfectly fine to proceed without one, and the performance difference will only become obvious with very large numbers of samples.">
                <options from_data_table="pangolin_assignment">
                    <column name="value" index="0" />
                    <column name="description" index="1" />
                    <column name="path" index="4" />
                    <filter type="static_value" column="2" value="@MIN_COMPATIBLE_PANGOLIN_DATA_FORMAT@" />
                    <filter type="param_value" ref="release" column="0" />
                </options>
            </param>
        </xml>
        <xml name="cached_pangolin_data">
            <when value="cached">
                <param name="release" label="Cached release of pangolin-data" type="select">
                    <options from_data_table="pangolin_data">
                        <column name="value" index="0" />
                        <column name="description" index="1" />
                        <column name="date" index="3" />
                        <column name="path" index="4" />
                        <filter type="sort_by" column="3" />
                        <filter type="static_value" column="2" value="@MIN_COMPATIBLE_PANGOLIN_DATA_FORMAT@" />
                        <validator type="no_options" message="No cached constellations release available" />
                    </options>
                </param>
                <yield />
            </when>
        </xml>
        <xml name="pangolin_data_sources">
            <conditional name="pangolin_data">
                <param name="source" type="select" label="Version of pangolin-data to use">
                    <option value="default">Use pangolin-data version (v@PANGOLIN_DATA_VERSION@) shipped with this version of the tool</option>
                    <option value="cached">Use specific pangolin-data version cached on this Galaxy server</option>
                    <option value="download">Download latest available pangolin-data version from web</option>
                </param>
                <when value="default" />
                <yield />
            </conditional>
        </xml>
    </macros>
    <requirements>
        <requirement type="package" version="@TOOL_VERSION@">pangolin</requirement>
        <!-- Pin also the versions of all core dependencies - the ones

        when updating the wrapper for other reasons. -->
        <requirement type="package" version="0.3.17">scorpio</requirement>
        <requirement type="package" version="@PANGOLIN_DATA_VERSION@">pangolin-data</requirement>
        <requirement type="package" version="@CONSTELLATIONS_VERSION@">constellations</requirement>
        <requirement type="package" version="0.6.2">usher</requirement>
        <requirement type="package" version="1.1.0">gofasta</requirement>
        <requirement type="package" version="426">ucsc-fatovcf</requirement>
        <requirement type="package" version="2.24">minimap2</requirement>
        <!-- wrapper-specific requirements to turn pangolin's native
        comma-separated output into tab-separated one and to truncate
        pangolin's all-versions output. -->
        <requirement type="package" version="0.25.0">csvtk</requirement>
        <requirement type="package" version="3.4">grep</requirement>
    </requirements>
    <version_command><![CDATA[pangolin --version]]></version_command>
    <command detect_errors="exit_code"><![CDATA[
## Prepare a pangolin datadir if required:
#if str($engine.pangolin_data.source) != 'default' or str($constellations.source) != 'default':

            --use-assignment-cache
        #end if
    #end if
#end if
#if $alignment:
    $alignment --alignment-file '$align1'
#end if
--outfile report.csv
--max-ambig $max_ambig
--min-length $min_length
$expanded_lineage

            help="The analysis engine to use for lineage assignment. Note: In this version, UShER mode is implemented through the usher-sampled algorithm of UShER. If you prefer to run the original usher algorithm, please use version 4.1.3 of the tool.">
                <option value="usher">UShER</option>
                <option value="pangolearn">pangoLEARN</option>
            </param>
            <when value="usher">
                <expand macro="pangolin_data_sources">
                    <expand macro="cached_pangolin_data">
                        <expand macro="cached_usher_assignment_cache" />
                    </expand>
                    <expand macro="usher_download_option" />
                </expand>
            </when>
            <when value="pangolearn">
                <expand macro="pangolin_data_sources">
                    <expand macro="cached_pangolin_data" />
                    <when value="download" />
                </expand>
            </when>
        </conditional>
        <conditional name="constellations">
            <param name="source" type="select" label="Version of constellations to use">
                <option value="default">Use constellations version (v@CONSTELLATIONS_VERSION@) shipped with this version of the tool</option>

                        <column name="description" index="1" />
                        <column name="date" index="3" />
                        <column name="path" index="4" />
                        <filter type="sort_by" column="3" />
                        <filter type="regexp" column="2" value="@COMPATIBLE_SCORPIO_DATA_FORMAT@" />
                        <validator type="no_options" message="No cached constellations release available" />
                    </options>
                </param>
            </when>
            <when value="download" />
        </conditional>

                        <action name="column_names" type="metadata" default="taxon,lineage,conflict,ambiguity_score,scorpio_call,scorpio_support,scorpio_conflict,scorpio_notes,version,pangolin_version,scorpio_version,constellation_version,is_designated,qc_status,qc_notes,note,expanded_lineage" />
                    </when>
                </conditional>
            </actions>
        </data>
        <data name="align1" format="fasta" label="pangolin alignment on ${on_string}">
            <filter>alignment</filter>
        </data>
    </outputs>
    <tests>
        <!-- Test only the default UShER mode for now since the

28:9d61d3fa046c

<tool id="pangolin" name="Pangolin" version="@TOOL_VERSION@+galaxy0" profile="20.01">
    <description>Phylogenetic Assignment of Outbreak Lineages</description>
    <macros>
        <token name="@TOOL_VERSION@">4.3</token>
        <token name="@PANGOLIN_DATA_VERSION@">1.21</token>
        <token name="@CONSTELLATIONS_VERSION@">0.1.12</token>
        <!-- a regex describing the pangolin-data versions that this wrapper version
        is backwards-compatible with; can be used with the min_pangolin_version
        column of the pangolin_data and pangolin_assignment data table to offer
        only compatible versions of that data. -->
        <token name="@MIN_COMPATIBLE_PANGOLIN_DATA_FORMAT@"><![CDATA[^4$|^4\.3$]]></token>
        <!-- pangolin-data versions >=1.20 do not contain pangoLEARN models anymore.
        This version of pangolin still supports pangolearn analysis mode, but only when using
        older versions of pangolin-data. -->
        <token name="@MIN_COMPATIBLE_PANGOLIN_DATA_FORMAT_FOR_PANGOLEARN@">4></token>
        <!-- a regex describing the scorpio versions that this wrapper version
        is backwards-compatible with; can be used with the min_scorpio_version
        column of the pangolin_constellations data table to offer only compatible
        versions of constellations data. -->
        <token name="@COMPATIBLE_SCORPIO_DATA_FORMAT@"><![CDATA[(^0\.[1-3]$|^0\.[0-2]\.\d+$|^0\.3\.\d$|^0\.3\.1[0-7]$|^0$)]]></token>
        <xml name="cached_usher_assignment_cache">
            <param name="assignment_cache_release" type="select" optional="true" label="Use corresponding UShER assignment cache?"
            help="If the server offers a copy of the UShER assignment cache along with the specified version of pangolin-data, you can select it here to speed up UShER lineage assignment. If no suitable assignment cache is available, it is perfectly fine to proceed without one, and the performance difference will only become obvious with very large numbers of samples.">
                <options from_data_table="pangolin_assignment">
                    <column name="value" index="0" />
                    <column name="description" index="1" />
                    <column name="path" index="4" />
                    <filter type="regexp" column="2" value="@MIN_COMPATIBLE_PANGOLIN_DATA_FORMAT@" />
                    <filter type="param_value" ref="release" column="0" />
                </options>
            </param>
        </xml>
    </macros>
    <requirements>
        <requirement type="package" version="@TOOL_VERSION@">pangolin</requirement>
        <!-- Pin also the versions of all core dependencies - the ones

        when updating the wrapper for other reasons. -->
        <requirement type="package" version="0.3.17">scorpio</requirement>
        <requirement type="package" version="@PANGOLIN_DATA_VERSION@">pangolin-data</requirement>
        <requirement type="package" version="@CONSTELLATIONS_VERSION@">constellations</requirement>
        <requirement type="package" version="0.6.2">usher</requirement>
        <requirement type="package" version="1.2.0">gofasta</requirement>
        <requirement type="package" version="448">ucsc-fatovcf</requirement>
        <requirement type="package" version="2.26">minimap2</requirement>
        <!-- wrapper-specific requirements to turn pangolin's native
        comma-separated output into tab-separated one and to truncate
        pangolin's all-versions output. -->
        <requirement type="package" version="0.26.0">csvtk</requirement>
        <requirement type="package" version="3.11">grep</requirement>
    </requirements>
    <version_command><![CDATA[pangolin --version]]></version_command>
    <command detect_errors="exit_code"><![CDATA[
## Prepare a pangolin datadir if required:
#if str($engine.pangolin_data.source) != 'default' or str($constellations.source) != 'default':

            --use-assignment-cache
        #end if
    #end if
#end if
#if $alignment:
    $alignment --alignment-file aln.fa
#end if
--outfile report.csv
--max-ambig $max_ambig
--min-length $min_length
$expanded_lineage

            help="The analysis engine to use for lineage assignment. Note: In this version, UShER mode is implemented through the usher-sampled algorithm of UShER. If you prefer to run the original usher algorithm, please use version 4.1.3 of the tool.">
                <option value="usher">UShER</option>
                <option value="pangolearn">pangoLEARN</option>
            </param>
            <when value="usher">
                <conditional name="pangolin_data">
                    <param name="source" type="select" label="Version of pangolin-data to use">
                        <option value="default">Use pangolin-data version (v@PANGOLIN_DATA_VERSION@) shipped with this version of the tool</option>
                        <option value="cached">Use specific pangolin-data version cached on this Galaxy server</option>
                        <option value="download">Download latest available pangolin-data version from web</option>
                    </param>
                    <when value="default" />
                    <when value="cached">
                        <param name="release" label="Cached release of pangolin-data" type="select">
                            <options from_data_table="pangolin_data">
                                <column name="value" index="0" />
                                <column name="description" index="1" />
                                <column name="date" index="3" />
                                <column name="path" index="4" />
                                <filter type="sort_by" column="3" />
                                <filter type="regexp" column="2" value="@MIN_COMPATIBLE_PANGOLIN_DATA_FORMAT@" />
                                <validator type="no_options" message="No compatible cached pangolin-data release available" />
                            </options>
                        </param>
                        <expand macro="cached_usher_assignment_cache" />
                    </when>
                    <when value="download">
                        <param argument="--use-assignment-cache" type="boolean" truevalue="--use-assignment-cache" falsevalue="" label="Download and use also latest UShER assignment cache?"
                        help="Get the latest UShER assignment cache from the pangolin-assignment online repository and use it to speed up UShER lineage assignment. Note: Downloading the cached assignments will only pay off for large numbers of input samples." />
                    </when>
                </conditional>
            </when>
            <when value="pangolearn">
                <conditional name="pangolin_data">
                    <param name="source" type="select" label="Version of pangolin-data to use"
                    help="Please note: pangoLEARN mode is considered deprecated and the underlying machine learning model has been removed from recent versions of pangolin-data. If you really still want to use this mode, you are restricted to the compatible historic pangolin-data versions cached on this Galaxy server.">
                        <option value="cached">Use specific pangolin-data version cached on this Galaxy server</option>
                    </param>
                    <when value="cached">
                        <param name="release" label="Cached release of pangolin-data" type="select">
                            <options from_data_table="pangolin_data">
                                <column name="value" index="0" />
                                <column name="description" index="1" />
                                <column name="date" index="3" />
                                <column name="path" index="4" />
                                <filter type="sort_by" column="3" />
                                <filter type="static_value" column="2" value="@MIN_COMPATIBLE_PANGOLIN_DATA_FORMAT_FOR_PANGOLEARN@" />
                                <validator type="no_options" message="No compatible cached pangolin-data release available" />
                            </options>
                        </param>
                    </when>
                </conditional>
            </when>
        </conditional>
        <conditional name="constellations">
            <param name="source" type="select" label="Version of constellations to use">
                <option value="default">Use constellations version (v@CONSTELLATIONS_VERSION@) shipped with this version of the tool</option>

                        <column name="description" index="1" />
                        <column name="date" index="3" />
                        <column name="path" index="4" />
                        <filter type="sort_by" column="3" />
                        <filter type="regexp" column="2" value="@COMPATIBLE_SCORPIO_DATA_FORMAT@" />
                        <validator type="no_options" message="No compatible cached constellations release available" />
                    </options>
                </param>
            </when>
            <when value="download" />
        </conditional>

                        <action name="column_names" type="metadata" default="taxon,lineage,conflict,ambiguity_score,scorpio_call,scorpio_support,scorpio_conflict,scorpio_notes,version,pangolin_version,scorpio_version,constellation_version,is_designated,qc_status,qc_notes,note,expanded_lineage" />
                    </when>
                </conditional>
            </actions>
        </data>
        <data name="align1" format="fasta" from_work_dir="aln.fa" label="pangolin alignment on ${on_string}">
            <filter>alignment</filter>
        </data>
    </outputs>
    <tests>
        <!-- Test only the default UShER mode for now since the