comparison augustus.xml @ 3:f5075dee9d6b draft

planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/augustus commit cf04d83d615ff09c4458982282d422fbef7d83ac

2:a77a4e9921e0

<tool id="augustus" name="Augustus" version="0.3.1">
    <description>gene prediction for eukaryotic genomes</description>
    <parallelism method="multi" split_inputs="input_genome" split_mode="to_size" split_size="10" merge_outputs="output,protein_output,codingseq_output"></parallelism>
    <requirements>
        <requirement type="package" version="2.7">augustus</requirement>
    </requirements>
    <stdio>
        <exit_code range="1:" />
        <exit_code range=":-1" />
        <regex match="Error:" />
        <regex match="Exception:" />
    </stdio>
    <command>
<![CDATA[
        ## please set export AUGUSTUS_CONFIG_PATH=/path_to_augustus/augustus/config
        ## or use the --AUGUSTUS_CONFIG_PATH=path if you are not installing through the toolshed
        ## Augustus writes the protein and coding sequences as comment into the gff/gtf file an external script is used to extract the sequences into additional files

        augustus
            --strand=$strand
            $noInFrameStop
            $gff

            #if 'protein' in str($outputs).split(','):
                --protein=on
            #else:
                --protein=off

                --cds=off
            #end if

            $singlestrand
            $input_genome
            $mea
            $utr
            --genemodel=$genemodel
            --species=$organism
            ##--outfile=$output
        | tee $output
        #if 'protein' in str($outputs).split(',') or 'codingseq' in str($outputs).split(','):
            | python $__tool_directory__/extract_features.py
                #if 'protein' in str($outputs).split(','):
                    --protein $protein_output

        #end if
]]>
    </command>
    <inputs>
        <param name="input_genome" type="data" format="fasta" label="Genome Sequence"/>
        <param name="noInFrameStop" type="boolean" label="Don't report transcripts with in-frame stop codons (--noInFrameStop)" truevalue="--noInFrameStop=true" falsevalue="--noInFrameStop=false" checked="false" help="Otherwise, intron-spanning stop codons could occur" />
        <param name="singlestrand" type="boolean" label="Predict genes independently on each strand, allow overlapping genes on opposite strands (--singlestrand)" truevalue="--singlestrand=true" falsevalue="--singlestrand=false" checked="false" />
        <param name="mea" type="boolean" label="Using the maximum expected accuracy approach (--mea)" truevalue="--mea=1" falsevalue="" checked="false" help="MEA is an alternative decoding approach." />
        <param name="utr" type="boolean" label="Predict the untranslated regions in addition to the coding sequence (--UTR)" truevalue="--UTR=on" falsevalue="--UTR=off" checked="false" help="This currently works only for human, galdieria, toxoplasma and caenorhabditis." />

        <param name="organism" label="Model Organism" type="select" multiple="false" format="text" help="Choose a specialised trainingset.">
            <option value="human">Homo sapiens</option>
            <option value="fly">Drosophila melanogaster</option>
            <option value="arabidopsis">Arabidopsis thaliana</option>

            <option value="tribolium2012">Tribolium castaneum</option>
            <option value="schistosoma">Schistosoma mansoni</option>
            <option value="tetrahymena">Tetrahymena thermophila</option>
            <option value="galdieria">Galdieria sulphuraria</option>
            <option value="maize">Zea mays</option>
            <option value="toxoplasma ">Toxoplasma gondii</option>
            <option value="caenorhabditis ">Caenorhabditis elegans</option>
            <option value="aspergillus_fumigatus">Aspergillus fumigatus</option>
            <option value="aspergillus_nidulans ">Aspergillus nidulans</option>
            <option value="aspergillus_oryzae ">Aspergillus oryzae</option>
            <option value="aspergillus_terreus">Aspergillus terreus</option>
            <option value="botrytis_cinerea ">Botrytis cinerea</option>
            <option value="candida_albicans ">Candida albicans</option>
            <option value="candida_guilliermondii ">Candida guilliermondii</option>
            <option value="candida_tropicalis ">Candida tropicalis</option>
            <option value="chaetomium_globosum">Chaetomium globosum</option>
            <option value="coccidioides_immitis ">Coccidioides immitis</option>
            <option value="coprinus ">Coprinus cinereus</option>
            <option value="coprinus_cinereus">Coprinus cinereus</option>
            <option value="cryptococcus_neoformans_gattii ">Cryptococcus neoformans gattii</option>
            <option value="cryptococcus_neoformans_neoformans_B ">Cryptococcus neoformans neoformans</option>
            <option value="cryptococcus_neoformans_neoformans_JEC21 ">Cryptococcus neoformans neoformans</option>
            <option value="debaryomyces_hansenii">Debaryomyces hansenii</option>
            <option value="encephalitozoon_cuniculi_GB">Encephalitozoon cuniculi</option>
            <option value="eremothecium_gossypii">Eremothecium gossypii</option>
            <option value="fusarium_graminearum ">Fusarium graminearum</option>
            <option value="histoplasma_capsulatum ">Histoplasma capsulatum</option>
            <option value="(histoplasma)">Histoplasma capsulatum</option>
            <option value="kluyveromyces_lactis ">Kluyveromyces lactis</option>
            <option value="laccaria_bicolor ">Laccaria bicolor</option>
            <option value="lamprey">Petromyzon marinus</option>
            <option value="leishmania_tarentolae">Leishmania tarentolae</option>
            <option value="lodderomyces_elongisporus">Lodderomyces elongisporus</option>
            <option value="magnaporthe_grisea ">Magnaporthe grisea</option>
            <option value="neurospora_crassa">Neurospora crassa</option>
            <option value="phanerochaete_chrysosporium">Phanerochaete chrysosporium</option>
            <option value="pichia_stipitis">Pichia stipitis</option>
            <option value="rhizopus_oryzae">Rhizopus oryzae</option>
            <option value="saccharomyces_cerevisiae_S288C ">Saccharomyces cerevisiae</option>
            <option value="saccharomyces_cerevisiae_rm11-1a_1 ">Saccharomyces cerevisiae</option>
            <option value="(saccharomyces)">Saccharomyces cerevisiae</option>
            <option value="schizosaccharomyces_pombe">Schizosaccharomyces pombe</option>
            <option value="trichinella">Trichinella spiralis</option>
            <option value="ustilago_maydis">Ustilago maydis</option>
            <option value="yarrowia_lipolytica">Yarrowia lipolytica</option>
            <option value="nasonia">Nasonia vitripennis</option>

            <option value="pneumocystis">Pneumocystis jirovecii</option>
            <option value="chicken">Gallus gallus domesticus (chicken)</option>
            <option value="cacao">Theobroma cacao (cacao)</option>
            <option value="heliconius_melpomene1">Heliconius melpomene</option>
            <option value="xenoturbella">Xenoturbella</option>
        </param>

        <param name="strand" type="select" multiple="false" format="text" help="Report predicted genes on both strands, just the forward or just the backward strand.">
            <option value="both">both</option>
            <option value="forward">forward</option>
            <option value="backward">backward</option>
        </param>

        <param name="genemodel" label="Gene Model" type="select" multiple="false" format="text" help="Gene Model to predict, for more information please refere to the help.">
            <option value="complete">complete</option>
            <option value="partial">partial</option>
            <option value="intronless">intronless</option>
            <option value="atleastone">atleastone</option>
            <option value="exactlyone">exactlyone</option>
            <option value="bacterium">bacterium (beta version)</option>
        </param>


        <param name="gff" type="boolean" label="GFF formated output, standard is GTF (--gff3)" truevalue="--gff3=on" falsevalue="--gff3=off" checked="false" />

        <param name="outputs" type="select" display="checkboxes" multiple="True" label="Output options">
            <option value="protein" selected="True">predicted protein sequences (--protein)</option>
            <option value="codingseq" selected="True">coding sequence as comment in the output file (--codingseq)</option>
            <option value="introns">predicted intron sequences (--introns)</option>

            <option value="cds" selected="true">CDS region (--cds)</option>

            <validator type="no_options" message="Please select at least one output file." />
        </param>


    </inputs>
    <outputs>
        <data format="gtf" name="output" label="${tool.name} on ${on_string}: GTF/GFF">
            <change_format>
                <when input="gff" value="--gff3=on" format="gff" />
            </change_format>
        </data>
        <data format="fasta" name="protein_output" label="${tool.name} on ${on_string}: Protein sequence">
            <filter>'protein' in outputs</filter>
        </data>

    </outputs>
    <tests>
        <test>
            <param name="input_genome" value="human_augustus.fa" ftype="fasta" />
            <param name="organism" value="human" />
            <param name="utr" value="--UTR=on" />
            <output name="output" file="human_augustus_utr-on.gtf" ftype="gtf" lines_diff="2"/>
        </test>
        <test>
            <param name="input_genome" value="human_augustus.fa" ftype="fasta" />
            <param name="organism" value="human" />
            <param name="utr" value="--UTR=on" />
            <param name="gff" value="--gff3=on" />
            <output name="output" file="human_augustus_utr-on.gff" ftype="gff3" lines_diff="2"/>
        </test>
        <test>
            <param name="input_genome" value="arabidopsis_augustus.fa" ftype="fasta" />
            <param name="organism" value="arabidopsis" />
            <param name="singlestrand" value="--singlestrand=true" />
            <param name="mea" value="--mea=1" />
            <output name="output" file="arabidopsis_augustus_utr-off_singlestrand-on_mea-on.gtf" ftype="gtf" lines_diff="2"/>
        </test>
        <test>
            <param name="input_genome" value="human_augustus.fa" ftype="fasta" />
            <param name="organism" value="human" />
            <param name="outputs" value="protein,codingseq,introns,cds" />
            <output name="output" file="human_augustus_protein_codingseq_introns_cds_main.gtf" ftype="gff" lines_diff="2"/>
            <output name="codingseq_output" file="human_augustus_protein_codingseq_introns_cds_codingseq.fasta" ftype="fasta" />
            <output name="protein_output" file="human_augustus_protein_codingseq_introns_cds_protein.fasta" ftype="fasta" />
        </test>
    </tests>
    <help>

3:f5075dee9d6b

<tool id="augustus" name="Augustus" version="3.1.0">
    <description>gene prediction for eukaryotic genomes</description>
    <parallelism method="multi" split_inputs="input_genome" split_mode="to_size" split_size="10" merge_outputs="output,protein_output,codingseq_output"></parallelism>
    <requirements>
        <requirement type="package" version="3.1">augustus</requirement>
    </requirements>
    <stdio>
        <exit_code range="1:" />
        <exit_code range=":-1" />
        <regex match="Error:" />
        <regex match="Exception:" />
    </stdio>
    <command>
<![CDATA[
        ## Augustus writes the protein and coding sequences as comment into the gff/gtf file an external
        ## script is used to extract the sequences into additional files

        augustus
            --strand=$strand
            $noInFrameStop
            $gff
            --uniqueGeneId=true

            #if 'protein' in str($outputs).split(','):
                --protein=on
            #else:
                --protein=off

                --cds=off
            #end if

            $singlestrand
            $input_genome
            $utr
            --genemodel=$genemodel
            --species=$organism

        | tee $output
        #if 'protein' in str($outputs).split(',') or 'codingseq' in str($outputs).split(','):
            | python $__tool_directory__/extract_features.py
                #if 'protein' in str($outputs).split(','):
                    --protein $protein_output

        #end if
]]>
    </command>
    <inputs>
        <param name="input_genome" type="data" format="fasta" label="Genome Sequence"/>
        <param name="noInFrameStop" argument="--noInFrameStop" type="boolean"
            label="Don't report transcripts with in-frame stop codons (--noInFrameStop)"
            truevalue="--noInFrameStop=true" falsevalue="--noInFrameStop=false" checked="false"
            help="Otherwise, intron-spanning stop codons could occur. (--noInFrameStop)" />
        <param name="singlestrand" type="boolean" argument="--singlestrand"
            label="Predict genes independently on each strand"
            help="This allows overlapping genes on opposite strands. (--singlestrand)"
            truevalue="--singlestrand=true" falsevalue="--singlestrand=false" checked="false" />

        <param name="utr" type="boolean" argument="--UTR"
            label="Predict the untranslated regions in addition to the coding sequence"
            truevalue="--UTR=on" falsevalue="--UTR=off" checked="false"
            help="This currently works only for human, galdieria, toxoplasma and caenorhabditis. (--UTR)" />

        <param name="organism" label="Model Organism" type="select" multiple="false" format="text" help="Choose a specialised trainingset.">
            <option value="human">Homo sapiens</option>
            <option value="fly">Drosophila melanogaster</option>
            <option value="arabidopsis">Arabidopsis thaliana</option>

            <option value="tribolium2012">Tribolium castaneum</option>
            <option value="schistosoma">Schistosoma mansoni</option>
            <option value="tetrahymena">Tetrahymena thermophila</option>
            <option value="galdieria">Galdieria sulphuraria</option>
            <option value="maize">Zea mays</option>
            <option value="toxoplasma">Toxoplasma gondii</option>
            <option value="caenorhabditis ">Caenorhabditis elegans</option>
            <option value="aspergillus_fumigatus">Aspergillus fumigatus</option>
            <option value="aspergillus_nidulans ">Aspergillus nidulans</option>
            <option value="aspergillus_oryzae ">Aspergillus oryzae</option>
            <option value="aspergillus_terreus">Aspergillus terreus</option>
            <option value="botrytis_cinerea ">Botrytis cinerea</option>
            <option value="candida_albicans ">Candida albicans</option>
            <option value="candida_guilliermondii ">Candida guilliermondii</option>
            <option value="candida_tropicalis ">Candida tropicalis</option>
            <option value="chaetomium_globosum">Chaetomium globosum</option>
            <option value="coccidioides_immitis">Coccidioides immitis</option>
            <option value="coprinus">Coprinus cinereus</option>
            <option value="coprinus_cinereus">Coprinus cinereus</option>
            <option value="cryptococcus_neoformans_gattii">Cryptococcus neoformans gattii</option>
            <option value="cryptococcus_neoformans_neoformans_B">Cryptococcus neoformans neoformans</option>
            <option value="cryptococcus_neoformans_neoformans_JEC21">Cryptococcus neoformans neoformans</option>
            <option value="cryptococcus">Cryptococcus neoformans</option>
            <option value="debaryomyces_hansenii">Debaryomyces hansenii</option>
            <option value="encephalitozoon_cuniculi_GB">Encephalitozoon cuniculi</option>
            <option value="eremothecium_gossypii">Eremothecium gossypii</option>
            <option value="fusarium_graminearum ">Fusarium graminearum</option>
            <option value="histoplasma_capsulatum ">Histoplasma capsulatum</option>
            <option value="histoplasma">Histoplasma capsulatum</option>
            <option value="kluyveromyces_lactis ">Kluyveromyces lactis</option>
            <option value="laccaria_bicolor ">Laccaria bicolor</option>
            <option value="lamprey">Petromyzon marinus</option>
            <option value="leishmania_tarentolae">Leishmania tarentolae</option>
            <option value="lodderomyces_elongisporus">Lodderomyces elongisporus</option>
            <option value="magnaporthe_grisea ">Magnaporthe grisea</option>
            <option value="neurospora_crassa">Neurospora crassa</option>
            <option value="phanerochaete_chrysosporium">Phanerochaete chrysosporium</option>
            <option value="pichia_stipitis">Pichia stipitis</option>
            <option value="rhizopus_oryzae">Rhizopus oryzae</option>
            <option value="saccharomyces_cerevisiae_S288C">Saccharomyces cerevisiae</option>
            <option value="saccharomyces_cerevisiae_rm11-1a_1">Saccharomyces cerevisiae</option>
            <option value="saccharomyces">Saccharomyces cerevisiae</option>
            <option value="schizosaccharomyces_pombe">Schizosaccharomyces pombe</option>
            <option value="trichinella">Trichinella spiralis</option>
            <option value="ustilago_maydis">Ustilago maydis</option>
            <option value="yarrowia_lipolytica">Yarrowia lipolytica</option>
            <option value="nasonia">Nasonia vitripennis</option>

            <option value="pneumocystis">Pneumocystis jirovecii</option>
            <option value="chicken">Gallus gallus domesticus (chicken)</option>
            <option value="cacao">Theobroma cacao (cacao)</option>
            <option value="heliconius_melpomene1">Heliconius melpomene</option>
            <option value="xenoturbella">Xenoturbella</option>
            <option value="E_coli_K12">E coli K12</option>
            <option value="c_elegans_trsk">c elegans trsk</option>
            <option value="camponotus_floridanus">Camponotus floridanus</option>
            <option value="coyote_tobacco">Coyote tobacco</option>
            <option value="s_aureus">Staphylococcus aureus</option>
            <option value="thermoanaerobacter_tengcongensis">Thermoanaerobacter tengcongensis</option>
            <option value="wheat">wheat</option>
            <option value="zebrafish">Danio rerio</option>
            <option value="anidulans">Aspergillus nidulans</option>
            <option value="bombus_impatiens1">Bombus impatiens1</option>
            <option value="bombus_terrestris2">Bombus terrestris2</option>
            <option value="botrytis_cinerea">Botrytis cinerea</option>
            <option value="brugia_malayi">Brugia malayi</option>
            <option value="conidiobolus_coronatus">Conidiobolus coronatus</option>
            <option value="cryptococcus_neoformans">Cryptococcus neoformans</option>
            <option value="culex_pipiens">Culex pipiens</option>
            <option value="elephant_shark">Callorhinchus milii</option>
            <option value="honeybee1">Apis mellifera</option>
            <option value="phanerochaete_chrysosporium">Phanerochaete chrysosporium</option>
            <option value="pea_aphid">Acyrthosiphon pisum</option>
            <option value="rhodnius_prolixus">Rhodnius prolixus</option>
            <option value="ustilago_maydis">Ustilago maydis</option>
            <option value="verticillium_albo_atrum1">Verticillium albo atrum1</option>
            <option value="verticillium_longisporum1">Verticillium longisporum1</option>
            <option value="Xipophorus_maculatus">Xipophorus_maculatus</option>
            <option value="adorsata">adorsata</option>
            <option value="ancylostoma_ceylanicum">ancylostoma_ceylanicum</option>
            <option value="maker2_athal1">maker2_athal1</option>
            <option value="maker2_c_elegans1">maker2_c_elegans1</option>
            <option value="maker2_dmel1">maker2_dmel1</option>
            <option value="maker2_spomb1">maker2_spomb1</option>
            <option value="parasteatoda">parasteatoda</option>
            <option value="rice">rice</option>
            <option value="schistosoma2">schistosoma2</option>
            <option value="sulfolobus_solfataricus">sulfolobus_solfataricus</option>
        </param> 

        <param name="strand" type="select" argument="--strand"
            label="Predict genes on specific strands" help="(--strand)">
            <option value="both">both</option>
            <option value="forward">forward</option>
            <option value="backward">backward</option>
        </param>

        <param name="genemodel" label="Gene Model" type="select"
            help="Gene Model to predict, for more information please refere to the help. (--genemodel)">
            <option value="complete">complete</option>
            <option value="partial">partial</option>
            <option value="intronless">intronless</option>
            <option value="atleastone">atleastone</option>
            <option value="exactlyone">exactlyone</option>
        </param>

        <param name="gff" type="boolean" label="GFF formated output"
            help="Standard output is GTF. (--gff3)"
            truevalue="--gff3=on" falsevalue="--gff3=off" checked="false" />

        <param name="outputs" type="select" display="checkboxes" multiple="True" label="Output options">
            <option value="protein" selected="True">predicted protein sequences (--protein)</option>
            <option value="codingseq" selected="True">coding sequence as comment in the output file (--codingseq)</option>
            <option value="introns">predicted intron sequences (--introns)</option>

            <option value="cds" selected="true">CDS region (--cds)</option>

            <validator type="no_options" message="Please select at least one output file." />
        </param>

    </inputs>
    <outputs>
        <data format="gtf" name="output" label="${tool.name} on ${on_string}: GTF/GFF">
            <change_format>
                <when input="gff" value="--gff3=on" format="gff3" />
            </change_format>
        </data>
        <data format="fasta" name="protein_output" label="${tool.name} on ${on_string}: Protein sequence">
            <filter>'protein' in outputs</filter>
        </data>

    </outputs>
    <tests>
        <test>
            <param name="input_genome" value="human_augustus.fa" ftype="fasta" />
            <param name="organism" value="human" />
            <param name="utr" value="True" />
            <output name="output" file="human_augustus_utr-on.gtf" ftype="gtf" lines_diff="6"/>
        </test>
        <test>
            <param name="input_genome" value="human_augustus.fa" ftype="fasta" />
            <param name="organism" value="human" />
            <param name="utr" value="True" />
            <param name="gff" value="True" />
            <output name="output" file="human_augustus_utr-on.gff" ftype="gff3" lines_diff="6"/>
        </test>
        <test>
            <param name="input_genome" value="human_augustus.fa" ftype="fasta" />
            <param name="organism" value="human" />
            <param name="outputs" value="protein,codingseq,introns,cds,start,stop" />
            <output name="output" file="human_augustus_protein_codingseq_introns_cds_main.gtf" ftype="gtf" lines_diff="6"/>
            <output name="codingseq_output" file="human_augustus_protein_codingseq_introns_cds_codingseq.fasta" ftype="fasta" />
            <output name="protein_output" file="human_augustus_protein_codingseq_introns_cds_protein.fasta" ftype="fasta" />
        </test>
    </tests>
    <help>