annotate phitk.xml @ 1:3a7f73d638ba draft default tip

planemo upload for repository https://github.com/Helmholtz-UFZ/ufz-galaxy-tools/blob/main/tools/phi-toolkit commit 368e8a7322c9763c648637263d4695abc146be13

<tool id="phi_toolkit_report" name="PHI toolkit report" version="@TOOL_VERSION@+galaxy@VERSION_SUFFIX@" profile="20.01" license="MIT">
    <description></description>
    <icon src="logo.jpg"/>
    <macros>
        <token name="@TOOL_VERSION@">0.2.0</token>
        <token name="@VERSION_SUFFIX@">0</token>

        <xml name="zenodo_collection_element" tokens="id" token_suffix="">
            <element name="@ID@@SUFFIX@" location="https://zenodo.org/records/15594463/files/@ID@.fasta.fasta?download=1" value="genomes/@ID@.fasta" ftype="fasta"/>
        </xml>

        <xml name="test_collection_element" tokens="id,folder,ext,suffix">
            <element name="@ID@@SUFFIX@" value="@FOLDER@/@ID@.fasta.@EXT@" ftype="@EXT@"/>
        </xml>
        <xml name="test_collection_non_empty" tokens="name,folder,ext" token_suffix="">
            <param name="@NAME@">
                <collection type="list">
                    <expand macro="test_collection_element" id="NC_000913" suffix="@SUFFIX@" folder="@FOLDER@" ext="@EXT@"/>
                    <expand macro="test_collection_element" id="NC_002737" suffix="@SUFFIX@" folder="@FOLDER@" ext="@EXT@"/>
                    <expand macro="test_collection_element" id="NC_008261" suffix="@SUFFIX@" folder="@FOLDER@" ext="@EXT@"/>
                    <expand macro="test_collection_element" id="NC_009012" suffix="@SUFFIX@" folder="@FOLDER@" ext="@EXT@"/>
                    <expand macro="test_collection_element" id="NC_014168" suffix="@SUFFIX@" folder="@FOLDER@" ext="@EXT@"/>
                    <expand macro="test_collection_element" id="NC_014212" suffix="@SUFFIX@" folder="@FOLDER@" ext="@EXT@"/>
                    <expand macro="test_collection_element" id="NC_014364" suffix="@SUFFIX@" folder="@FOLDER@" ext="@EXT@"/>
                    <expand macro="test_collection_element" id="NC_015761" suffix="@SUFFIX@" folder="@FOLDER@" ext="@EXT@"/>
                    <expand macro="test_collection_element" id="NC_017033" suffix="@SUFFIX@" folder="@FOLDER@" ext="@EXT@"/>
                    <expand macro="test_collection_element" id="NC_018068" suffix="@SUFFIX@" folder="@FOLDER@" ext="@EXT@"/>
                    <expand macro="test_collection_element" id="NC_018515" suffix="@SUFFIX@" folder="@FOLDER@" ext="@EXT@"/>
                    <expand macro="test_collection_element" id="NC_019936" suffix="@SUFFIX@" folder="@FOLDER@" ext="@EXT@"/>
                    <expand macro="test_collection_element" id="NC_021184" suffix="@SUFFIX@" folder="@FOLDER@" ext="@EXT@"/>
                    <yield/>
                </collection>
            </param>
        </xml>
        <xml name="test_collection" tokens="name,folder,ext" token_suffix="">
            <expand macro="test_collection_non_empty" name="@NAME@" folder="@FOLDER@" ext="@EXT@" suffix="@SUFFIX@">
                <expand macro="test_collection_element" id="NC_003450" suffix="@SUFFIX@" folder="@FOLDER@" ext="@EXT@"/>
                <expand macro="test_collection_element" id="NC_012982" suffix="@SUFFIX@" folder="@FOLDER@" ext="@EXT@"/>
                <expand macro="test_collection_element" id="NC_014008" suffix="@SUFFIX@" folder="@FOLDER@" ext="@EXT@"/>
                <expand macro="test_collection_element" id="NC_014211" suffix="@SUFFIX@" folder="@FOLDER@" ext="@EXT@"/>
                <expand macro="test_collection_element" id="NC_014363" suffix="@SUFFIX@" folder="@FOLDER@" ext="@EXT@"/>
                <expand macro="test_collection_element" id="NC_017095" suffix="@SUFFIX@" folder="@FOLDER@" ext="@EXT@"/>
                <expand macro="test_collection_element" id="NC_018014" suffix="@SUFFIX@" folder="@FOLDER@" ext="@EXT@"/>
                <expand macro="test_collection_element" id="NC_019897" suffix="@SUFFIX@" folder="@FOLDER@" ext="@EXT@"/>
                <expand macro="test_collection_element" id="NC_019904" suffix="@SUFFIX@" folder="@FOLDER@" ext="@EXT@"/>
            </expand>
        </xml>
        
    </macros>
    <!-- TODO <xrefs>
        <xref type="bio.tools"></xref>
    </xrefs> -->
    <requirements>
        <requirement type="package" version="2.74.0">bioconductor-biostrings</requirement>
        <requirement type="package" version="1.58.0">bioconductor-genomicranges</requirement>
        <requirement type="package" version="1.18.0">bioconductor-gmoviz</requirement>
        <requirement type="package" version="0.2.1">r-formattable</requirement>
        <requirement type="package" version="1.0.1">r-here</requirement>
        <requirement type="package" version="2.2.1">r-janitor</requirement>
        <requirement type="package" version="1.4.0">r-kableextra</requirement>
        <requirement type="package" version="1.3.0">r-patchwork</requirement>
        <requirement type="package" version="2.0.0">r-tidyverse</requirement>
        <requirement type="package" version="2.0.2">r-base64</requirement>
        <requirement type="package" version="3.5.0">r-pdftools</requirement>
    </requirements>
    <version_command><![CDATA[
        echo $(R --vanilla --slave -e 'required_packages <- c("tidyverse", "janitor", "here", "kableExtra", "gmoviz", "circlize", "GenomicRanges", "patchwork", "fs", "tools", "scales", "formattable", "pdftools", "base64"); invisible(lapply(required_packages, library, character.only = TRUE, quietly=TRUE, verbose=FALSE)); sessionInfo()' 2> /dev/null)
    ]]></version_command>
    <command detect_errors="exit_code"><![CDATA[
        #import re
        mkdir -p data/genomes &&
        #for e in $genomes
            #set id=re.sub(r'[^\w\-.]', '_', $e.element_identifier)
            ln -s '$e' 'data/genomes/${id}.fna' &&
        #end for

        mkdir -p data/host_analyses &&
        mkdir -p data/virus_analyses &&

        ## checkm
        mkdir -p data/host_analyses/checkm2 &&
        ln -s '$checkm' data/host_analyses/checkm2/quality_report.tsv &&

        ## defense-finder inputs
        mkdir -p data/host_analyses/defense-finder &&
        #for e in $defense_finder
            #set id=re.sub(r'[^\w\-.]', '_', $e.element_identifier)
            mkdir -p 'data/host_analyses/defense-finder/${id}/' &&
            ln -s '$e' 'data/host_analyses/defense-finder/${id}/${id}_defense_finder_systems.tsv' &&
        #end for

        ## genomad inputs
        mkdir -p data/host_analyses/genomad &&
        #for e in $genomad
            #set id=re.sub(r'[^\w\-.]', '_', $e.element_identifier)
            mkdir -p 'data/host_analyses/genomad/${id}/${id}_summary/' &&
            ln -s '$e' 'data/host_analyses/genomad/${id}/${id}_summary/${id}_virus_summary.tsv' &&
        #end for
        #for e in $genomad_phages
            #set id=re.sub(r'[^\w\-.]', '_', $e.element_identifier)
            mkdir -p 'data/host_analyses/genomad/${id}/${id}_summary/' &&
            ln -s '$e' 'data/host_analyses/genomad/${id}/${id}_summary/${id}_virus.fna' &&
        #end for
        #for e in $genomad_annotations
            #set id=re.sub(r'[^\w\-.]', '_', $e.element_identifier)
            mkdir -p 'data/host_analyses/genomad/${id}/${id}_summary/' &&
            ln -s '$e' 'data/host_analyses/genomad/${id}/${id}_summary/${id}_virus_genes.tsv' &&
        #end for

        ## GTDBTK
        mkdir -p data/host_analyses/gtdbtk &&
        ln -s '$gtdb_summary' data/host_analyses/gtdbtk/gtdbtk.bac120.summary.tsv &&

        ## CheckV
        mkdir -p data/virus_analyses/checkv &&
        #for e in $checkv
            #set id=re.sub(r'[^\w\-.]', '_', $e.element_identifier)
            mkdir -p 'data/virus_analyses/checkv/${id}/' &&
            ln -s '$e' 'data/virus_analyses/checkv/${id}/quality_summary.tsv' &&
        #end for

        ## drep compare
        mkdir -p data/virus_analyses/drep_compare &&
        #for e in $drep_compare
            #set id=re.sub(r'[^\w\-.]', '_', $e.element_identifier)
            mkdir -p 'data/virus_analyses/drep_compare/${id}/data_tables' &&
            ln -s '$e' 'data/virus_analyses/drep_compare/${id}/data_tables/Cdb.csv' &&
        #end for
        #for e in $drep_compare_clustering_dendrogram
            #set id=re.sub(r'[^\w\-.]', '_', $e.element_identifier)
            mkdir 'data/virus_analyses/drep_compare/${id}/figures/' &&
            ln -s '$e' 'data/virus_analyses/drep_compare/${id}/figures/Primary_clustering_dendrogram.pdf' &&
        #end for

        ## iphop
        mkdir -p data/virus_analyses/iphop &&
        #for e in $iphop
            #set id=re.sub(r'[^\w\-.]', '_', $e.element_identifier)
            mkdir -p 'data/virus_analyses/iphop/${id}/' &&
            ln -s '$e' 'data/virus_analyses/iphop/${id}/Host_prediction_to_genome_m90.csv' &&
        #end for

        ## abricate
        mkdir -p data/virus_analyses/abricate &&
        #for e in $abricate
            #set id=re.sub(r'[^\w\-.]', '_', $e.element_identifier)
            mkdir -p 'data/virus_analyses/abricate/${id}/' &&
            ln -s '$e' 'data/virus_analyses/abricate/${id}/${id}_virus_vfdb.tsv' &&
        #end for

        ## vibrant
        mkdir -p data/virus_analyses/vibrant &&
        #for e in $vibrant
            #set id=re.sub(r'[^\w\-.]', '_', $e.element_identifier)
            mkdir -p 'data/virus_analyses/vibrant/${id}/VIBRANT_${id}_virus/VIBRANT_results_${id}_virus/' &&
            ln -s '$e' 'data/virus_analyses/vibrant/${id}/VIBRANT_${id}_virus/VIBRANT_results_${id}_virus/VIBRANT_AMG_individuals_${id}_virus.tsv' &&
        #end for

        Rscript -e "library(rmarkdown); render('$__tool_directory__/report.Rmd', output_dir='"\$(pwd)"', output_format = 'html_document', intermediates_dir = '"\$(pwd)"', knit_root_dir='"\$(pwd)"')" &&
        cp report.html '$report'
        && >&2 echo "debug.log:"
        && >&2 cat debug.log
    ]]></command>
    <inputs>
        <param type="data_collection" name="genomes" format="fasta" label="Genomes"/>
        <param type="data" name="checkm" format="tabular" label="Checkm2 quality report"/>
        <param type="data_collection" name="genomad" format="tabular" label="geNomad Virus Summary"/>
        <param type="data_collection" name="genomad_phages" format="fasta" label="geNomad Virus fasta"/>
        <param type="data_collection" name="genomad_annotations" format="tabular" label="geNomad Virus genes"/>
        <param type="data_collection" name="defense_finder" format="tabular" label="DefenseFinder systems"/>
        <param type="data" name="gtdb_summary" format="tsv" label="GTDB-Tk summary"/>
        <param type="data_collection" name="checkv" format="tabular" label="CheckV Quality summary"/> 
        <param type="data_collection" name="drep_compare" format="csv" label="drep compare Cdb.csv"/>
        <param type="data_collection" name="drep_compare_clustering_dendrogram" format="pdf" label="drep compare clustering dendrogram"/>
        <param type="data_collection" name="iphop" format="csv" label="iPHop Host prediction to genome"/>
        <param type="data_collection" name="abricate" format="tabular" label="Abricate results"/> 
        <param type="data_collection" name="vibrant" format="tabular" label="Vibrant Individual predicted virus AMGs" help="Vibrant output option 'Individual predicted virus AMGs by protein and its respective genome'"/>
    </inputs>
    <outputs>
        <data name="report" format="html"/>
    </outputs>
    <tests>
        <test>
            <param name="genomes">
                <collection type="list">
                    <expand macro="zenodo_collection_element" id="NC_000913"/>
                    <expand macro="zenodo_collection_element" id="NC_002737"/>
                    <expand macro="zenodo_collection_element" id="NC_008261"/>
                    <expand macro="zenodo_collection_element" id="NC_009012"/>
                    <expand macro="zenodo_collection_element" id="NC_014168"/>
                    <expand macro="zenodo_collection_element" id="NC_014212"/>
                    <expand macro="zenodo_collection_element" id="NC_014364"/>
                    <expand macro="zenodo_collection_element" id="NC_015761"/>
                    <expand macro="zenodo_collection_element" id="NC_017033"/>
                    <expand macro="zenodo_collection_element" id="NC_018068"/>
                    <expand macro="zenodo_collection_element" id="NC_018515"/>
                    <expand macro="zenodo_collection_element" id="NC_019936"/>
                    <expand macro="zenodo_collection_element" id="NC_021184"/>
                    <expand macro="zenodo_collection_element" id="NC_003450"/>
                    <expand macro="zenodo_collection_element" id="NC_012982"/>
                    <expand macro="zenodo_collection_element" id="NC_014008"/>
                    <expand macro="zenodo_collection_element" id="NC_014211"/>
                    <expand macro="zenodo_collection_element" id="NC_014363"/>
                    <expand macro="zenodo_collection_element" id="NC_017095"/>
                    <expand macro="zenodo_collection_element" id="NC_018014"/>
                    <expand macro="zenodo_collection_element" id="NC_019897"/>
                    <expand macro="zenodo_collection_element" id="NC_019904"/>
                </collection>
            </param>
            <param name="checkm" value="checkm2.Quality_report.tabular"/>
            <expand macro="test_collection" name="genomad" folder="genomad_virus_summary" ext="tabular"/>
            <expand macro="test_collection" name="genomad_phages" folder="genomad_virus_fasta" ext="fasta"/>
            <expand macro="test_collection" name="genomad_annotations" folder="genomad_virus_genes" ext="tabular"/>
            <expand macro="test_collection" name="defense_finder" folder="defense_finder_systems" ext="tabular"/>
            <param name="gtdb_summary" value="gtdbtk.bac120.summary"/>
            <expand macro="test_collection_non_empty" name="checkv" folder="checkv_quality_summary" ext="tabular"/>
            <expand macro="test_collection_non_empty" name="drep_compare" folder="drep_compare" ext="csv"/>
            <expand macro="test_collection_non_empty" name="drep_compare_clustering_dendrogram" folder="drep_compare_clustering_dendrogram" ext="pdf"/>
            <expand macro="test_collection_non_empty" name="iphop" folder="iphop_host_prediction_to_genome" ext="csv"/> 
            <expand macro="test_collection_non_empty" name="abricate" folder="abricate_results" ext="tabular"/> 
            <expand macro="test_collection_non_empty" name="vibrant" folder="vibrant_amg_individuals" ext="tabular"/> 
            <output name="report">
                <assert_contents>
                    <has_size min="100"/>
                    <has_line line="&lt;html&gt;"/>
                    <has_text text="Error: " negate="true"/>
                </assert_contents>
            </output>
        </test>

        <!-- test with collection element identifiers contain suffix .fasta
             also GTDBTk and CheckM2 output tables contain the .fasta suffix-->
        <test>
            <param name="genomes">
                <collection type="list">
                    <expand macro="zenodo_collection_element" id="NC_000913" suffix=".fasta"/>
                    <expand macro="zenodo_collection_element" id="NC_002737" suffix=".fasta"/>
                    <expand macro="zenodo_collection_element" id="NC_008261" suffix=".fasta"/>
                    <expand macro="zenodo_collection_element" id="NC_009012" suffix=".fasta"/>
                    <expand macro="zenodo_collection_element" id="NC_014168" suffix=".fasta"/>
                    <expand macro="zenodo_collection_element" id="NC_014212" suffix=".fasta"/>
                    <expand macro="zenodo_collection_element" id="NC_014364" suffix=".fasta"/>
                    <expand macro="zenodo_collection_element" id="NC_015761" suffix=".fasta"/>
                    <expand macro="zenodo_collection_element" id="NC_017033" suffix=".fasta"/>
                    <expand macro="zenodo_collection_element" id="NC_018068" suffix=".fasta"/>
                    <expand macro="zenodo_collection_element" id="NC_018515" suffix=".fasta"/>
                    <expand macro="zenodo_collection_element" id="NC_019936" suffix=".fasta"/>
                    <expand macro="zenodo_collection_element" id="NC_021184" suffix=".fasta"/>
                    <expand macro="zenodo_collection_element" id="NC_003450" suffix=".fasta"/>
                    <expand macro="zenodo_collection_element" id="NC_012982" suffix=".fasta"/>
                    <expand macro="zenodo_collection_element" id="NC_014008" suffix=".fasta"/>
                    <expand macro="zenodo_collection_element" id="NC_014211" suffix=".fasta"/>
                    <expand macro="zenodo_collection_element" id="NC_014363" suffix=".fasta"/>
                    <expand macro="zenodo_collection_element" id="NC_017095" suffix=".fasta"/>
                    <expand macro="zenodo_collection_element" id="NC_018014" suffix=".fasta"/>
                    <expand macro="zenodo_collection_element" id="NC_019897" suffix=".fasta"/>
                    <expand macro="zenodo_collection_element" id="NC_019904" suffix=".fasta"/>
                </collection>
            </param>
            <param name="checkm" value="checkm2.Quality_report-wext.tabular"/>
            <expand macro="test_collection" name="genomad" suffix=".fasta" folder="genomad_virus_summary" ext="tabular"/>
            <expand macro="test_collection" name="genomad_phages" suffix=".fasta" folder="genomad_virus_fasta" ext="fasta"/>
            <expand macro="test_collection" name="genomad_annotations" suffix=".fasta" folder="genomad_virus_genes" ext="tabular"/>
            <expand macro="test_collection" name="defense_finder" suffix=".fasta" folder="defense_finder_systems" ext="tabular"/>
            <param name="gtdb_summary" value="gtdbtk.bac120-wext.summary"/>
            <expand macro="test_collection_non_empty" name="checkv" suffix=".fasta" folder="checkv_quality_summary" ext="tabular"/>
            <expand macro="test_collection_non_empty" name="drep_compare" suffix=".fasta" folder="drep_compare" ext="csv"/>
            <expand macro="test_collection_non_empty" name="drep_compare_clustering_dendrogram" suffix=".fasta" folder="drep_compare_clustering_dendrogram" ext="pdf"/>
            <expand macro="test_collection_non_empty" name="iphop" suffix=".fasta" folder="iphop_host_prediction_to_genome" ext="csv"/> 
            <!-- TODO remove from test-data <expand macro="test_collection_non_empty" name="phatyp" folder="phabox_phatyp" ext="tabular"/>  -->
            <expand macro="test_collection_non_empty" name="abricate" suffix=".fasta" folder="abricate_results" ext="tabular"/> 
            <expand macro="test_collection_non_empty" name="vibrant" suffix=".fasta" folder="vibrant_amg_individuals" ext="tabular"/> 
            <output name="report">
                <assert_contents>
                    <has_size min="100"/>
                    <has_line line="&lt;html&gt;"/>
                    <has_text text="Error: " negate="true"/>
                </assert_contents>
            </output>
        </test>
    </tests>
    <help><![CDATA[

.. class:: infomark

**What it does**

Create a report for the PHI toolkit workflow.

Usage
.....

**Input**

- Genomes
- Checkm2 quality report
- geNomad Virus Summary
- geNomad Virus genes
- DefenseFinder systems
- GTDB-Tk summary
- CheckV Quality summary
- drep compare Cdb.csv
- drep compare clustering dendrogram
- iPHop Host prediction to genome
- Abricate results
- Vibrant Individual predicted virus AMGs

**Output**

A html report summarizing the results.
    ]]></help>
    <!-- <citations>
        <citation type="doi"> </citation>
    </citations> -->
</tool>