comparison orthofinder_only_groups.xml @ 9:b3d25fae5389 draft default tip

planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/orthofinder commit 179e3fd2aeb66484e49e6370e9507e24d118f4ea

8:3b974afde673

<tool name="OrthoFinder" id="orthofinder_onlygroups" version="@TOOL_VERSION@+galaxy@VERSION_SUFFIX@">
    <description>finds orthogroups in a set of proteomes</description>
    <xrefs>
        <xref type="bio.tools">OrthoFinder</xref>
    </xrefs>
    <macros>
        <import>macros.xml</import>
    </macros>
    <requirements>
        <requirement type="package" version="@TOOL_VERSION@">orthofinder</requirement>
        <requirement type="package" version="2.36">util-linux</requirement>
    </requirements>
    <command detect_errors="exit_code"><![CDATA[
        #import re
        ## prepare inputs
        #if $init.start == "fasta":

    </inputs>
    <outputs>
        <!-- Orthogroups results -->
        <data format="txt" name="orthogroups1" label="OrthoFinder on ${on_string}: orthogroups (txt)" from_work_dir="results/Orthogroups/Orthogroups.txt" />
        <data format="tsv" name="orthogroups2" label="OrthoFinder on ${on_string}: orthogroups (tsv)" from_work_dir="results/Orthogroups/Orthogroups.tsv" />
        <data format="tsv" name="specs_overlap" label="OrthoFinder on ${on_string}: species overlaps" from_work_dir="results/Comparative_Genomics_Statistics/Orthogroups_SpeciesOverlaps.tsv" />
        <data format="tsv" name="unassigned_genes" label="OrthoFinder on ${on_string}: unassigned genes" from_work_dir="results/Orthogroups/Orthogroups_UnassignedGenes.tsv" />
        <data format="tsv" name="stat_overall" label="OrthoFinder on ${on_string}: overall comparative genomics statistics" from_work_dir="results/Comparative_Genomics_Statistics/Statistics_Overall.tsv" />
        <data format="tsv" name="stat_specs" label="OrthoFinder on ${on_string}: per species comparative genomics statistics" from_work_dir="results/Comparative_Genomics_Statistics/Statistics_PerSpecies.tsv" />

            <output name="unassigned_genes" value="results_fromblast/Orthogroups_UnassignedGenes.tsv"/>
            <output name="stat_overall" value="results_fromblast/Statistics_Overall.tsv" lines_diff="2"/>
            <output name="stat_specs" value="results_fromblast/Statistics_PerSpecies.tsv"/>
        </test>
        <!-- full mode + diamond + input files have extension fasta/faa/fa -->
        <test expect_num_outputs="9">
            <conditional name="init">
                <param name="start" value="fasta" />
                <param name="input_fasta" ftype="fasta" value="inputs/proteomes/Mycoplasma_agalactiae.faa,inputs/proteomes/Mycoplasma_gallisepticum.faa,inputs/proteomes/Mycoplasma_genitalium.faa,inputs/proteomes/Mycoplasma_hyopneumoniae.faa" />
                <conditional name="search">
                    <param name="search_program" value="diamond"/>

                    <has_n_columns n="5"/>
                </assert_contents>
            </output>
            <output name="species_tree" value="results/SpeciesTree_rooted.txt" compare="sim_size"/>
            <output_collection name="genetrees" type="list" count="0"/>
        </test>
        <!-- full mode + diamond + input files have extension fasta/faa/fa + duplications -->
        <test expect_num_outputs="14">
            <conditional name="init">
                <param name="start" value="fasta" />
                <param name="input_fasta" ftype="fasta" value="inputs/proteomes/Mycoplasma_agalactiae.faa,inputs/proteomes/Mycoplasma_gallisepticum.faa,inputs/proteomes/Mycoplasma_genitalium.faa,inputs/proteomes/Mycoplasma_hyopneumoniae.faa" />
                <conditional name="search">
                    <param name="search_program" value="diamond"/>

            <output name="duplications" value="results/Duplications.tsv" compare="sim_size"/>
            <output name="duplications_per_orthogroup" value="results/Duplications_per_Orthogroup.tsv" compare="sim_size"/>
            <output name="duplications_per_species_tree_node" value="results/Duplications_per_Species_Tree_Node.tsv" compare="sim_size"/>
            <output_collection name="genetrees" type="list" count="325"/>
            <output_collection name="resolved_trees" type="list" count="325"/>
        </test>
        <!-- trees + diamond + input files have no extension fasta/faa/fa + msa -->
        <test expect_num_outputs="6">
            <conditional name="init">
                <param name="start" value="fasta" />

                    <has_n_columns n="5"/>
                </assert_contents>
            </output>
        </test>
        <!-- trees + diamond + input files have no extension fasta/faa/fa + msa -->
        <test expect_num_outputs="9">
            <conditional name="init">
                <param name="start" value="fasta" />
                <param name="input_fasta" ftype="fasta" value="inputs/proteomes/Mycoplasma_agalactiae.faa,inputs/proteomes/Mycoplasma_gallisepticum.faa,inputs/proteomes/Mycoplasma_genitalium.faa,inputs/proteomes/Mycoplasma_hyopneumoniae.faa" />
                <conditional name="search">
                    <param name="search_program" value="diamond"/>

                    <has_text text="Number of genes per-species in orthogroup"/>
                    <has_text text="Percentage of orthogroups"/>
                    <has_text text="Number of orthogroups"/>
                    <has_text text="Number of genes"/>
                    <has_n_columns n="5"/>
                </assert_contents>
            </output>
            <assert_command>
                <has_text text="-M 'msa'"/>
                <has_text text="-A 'muscle'"/>

9:b3d25fae5389

<tool name="OrthoFinder" id="orthofinder_onlygroups" version="@TOOL_VERSION@+galaxy@VERSION_SUFFIX@">
    <description>finds orthogroups in a set of proteomes</description>
    <macros>
        <import>macros.xml</import>
    </macros>
    <xrefs>
        <xref type="bio.tools">OrthoFinder</xref>
    </xrefs>
    <requirements>
        <requirement type="package" version="@TOOL_VERSION@">orthofinder</requirement>
    </requirements>
    <command detect_errors="exit_code"><![CDATA[
        #import re
        ## prepare inputs
        #if $init.start == "fasta":

    </inputs>
    <outputs>
        <!-- Orthogroups results -->
        <data format="txt" name="orthogroups1" label="OrthoFinder on ${on_string}: orthogroups (txt)" from_work_dir="results/Orthogroups/Orthogroups.txt" />
        <data format="tsv" name="orthogroups2" label="OrthoFinder on ${on_string}: orthogroups (tsv)" from_work_dir="results/Orthogroups/Orthogroups.tsv" />
        <data format="tsv" name="hogs" label="OrthoFinder on ${on_string}: hierarchical orthogroups (tsv)" from_work_dir="results/Phylogenetic_Hierarchical_Orthogroups/N0.tsv" >
            <filter>trees['run_mode'] == "full"</filter>
        </data>
        <data format="tsv" name="specs_overlap" label="OrthoFinder on ${on_string}: species overlaps" from_work_dir="results/Comparative_Genomics_Statistics/Orthogroups_SpeciesOverlaps.tsv" />
        <data format="tsv" name="unassigned_genes" label="OrthoFinder on ${on_string}: unassigned genes" from_work_dir="results/Orthogroups/Orthogroups_UnassignedGenes.tsv" />
        <data format="tsv" name="stat_overall" label="OrthoFinder on ${on_string}: overall comparative genomics statistics" from_work_dir="results/Comparative_Genomics_Statistics/Statistics_Overall.tsv" />
        <data format="tsv" name="stat_specs" label="OrthoFinder on ${on_string}: per species comparative genomics statistics" from_work_dir="results/Comparative_Genomics_Statistics/Statistics_PerSpecies.tsv" />

            <output name="unassigned_genes" value="results_fromblast/Orthogroups_UnassignedGenes.tsv"/>
            <output name="stat_overall" value="results_fromblast/Statistics_Overall.tsv" lines_diff="2"/>
            <output name="stat_specs" value="results_fromblast/Statistics_PerSpecies.tsv"/>
        </test>
        <!-- full mode + diamond + input files have extension fasta/faa/fa -->
        <test expect_num_outputs="10">
            <conditional name="init">
                <param name="start" value="fasta" />
                <param name="input_fasta" ftype="fasta" value="inputs/proteomes/Mycoplasma_agalactiae.faa,inputs/proteomes/Mycoplasma_gallisepticum.faa,inputs/proteomes/Mycoplasma_genitalium.faa,inputs/proteomes/Mycoplasma_hyopneumoniae.faa" />
                <conditional name="search">
                    <param name="search_program" value="diamond"/>

                    <has_n_columns n="5"/>
                </assert_contents>
            </output>
            <output name="species_tree" value="results/SpeciesTree_rooted.txt" compare="sim_size"/>
            <output_collection name="genetrees" type="list" count="0"/>
            <output name="hogs">
                <assert_contents>
                    <has_text text="HOG" />
                    <has_text text="OG" />
                    <has_text text="Gene Tree Parent Clade" />
                    <has_text text="Mycoplasma_agalactiae"/>
                    <has_text text="Mycoplasma_gallisepticum"/>
                    <has_text text="Mycoplasma_genitalium"/>
                    <has_text text="Mycoplasma_hyopneumoniae"/>
                    <has_n_columns n="7"/>
                </assert_contents>
            </output>
        </test>
        <!-- full mode + diamond + input files have extension fasta/faa/fa + duplications -->
        <test expect_num_outputs="15">
            <conditional name="init">
                <param name="start" value="fasta" />
                <param name="input_fasta" ftype="fasta" value="inputs/proteomes/Mycoplasma_agalactiae.faa,inputs/proteomes/Mycoplasma_gallisepticum.faa,inputs/proteomes/Mycoplasma_genitalium.faa,inputs/proteomes/Mycoplasma_hyopneumoniae.faa" />
                <conditional name="search">
                    <param name="search_program" value="diamond"/>

            <output name="duplications" value="results/Duplications.tsv" compare="sim_size"/>
            <output name="duplications_per_orthogroup" value="results/Duplications_per_Orthogroup.tsv" compare="sim_size"/>
            <output name="duplications_per_species_tree_node" value="results/Duplications_per_Species_Tree_Node.tsv" compare="sim_size"/>
            <output_collection name="genetrees" type="list" count="325"/>
            <output_collection name="resolved_trees" type="list" count="325"/>
            <output name="hogs">
                <assert_contents>
                    <has_text text="HOG" />
                    <has_text text="OG" />
                    <has_text text="Gene Tree Parent Clade" />
                    <has_text text="Mycoplasma_agalactiae"/>
                    <has_text text="Mycoplasma_gallisepticum"/>
                    <has_text text="Mycoplasma_genitalium"/>
                    <has_text text="Mycoplasma_hyopneumoniae"/>
                    <has_n_columns n="7"/>
                </assert_contents>
            </output>
        </test>
        <!-- trees + diamond + input files have no extension fasta/faa/fa + msa -->
        <test expect_num_outputs="6">
            <conditional name="init">
                <param name="start" value="fasta" />

                    <has_n_columns n="5"/>
                </assert_contents>
            </output>
        </test>
        <!-- trees + diamond + input files have no extension fasta/faa/fa + msa -->
        <test expect_num_outputs="10">
            <conditional name="init">
                <param name="start" value="fasta" />
                <param name="input_fasta" ftype="fasta" value="inputs/proteomes/Mycoplasma_agalactiae.faa,inputs/proteomes/Mycoplasma_gallisepticum.faa,inputs/proteomes/Mycoplasma_genitalium.faa,inputs/proteomes/Mycoplasma_hyopneumoniae.faa" />
                <conditional name="search">
                    <param name="search_program" value="diamond"/>

                    <has_text text="Number of genes per-species in orthogroup"/>
                    <has_text text="Percentage of orthogroups"/>
                    <has_text text="Number of orthogroups"/>
                    <has_text text="Number of genes"/>
                    <has_n_columns n="5"/>
                </assert_contents>
            </output>
            <output name="hogs">
                <assert_contents>
                    <has_text text="HOG" />
                    <has_text text="OG" />
                    <has_text text="Gene Tree Parent Clade" />
                    <has_text text="Mycoplasma_agalactiae"/>
                    <has_text text="Mycoplasma_gallisepticum"/>
                    <has_text text="Mycoplasma_genitalium"/>
                    <has_text text="Mycoplasma_hyopneumoniae"/>
                    <has_n_columns n="7"/>
                </assert_contents>
            </output>
            <assert_command>
                <has_text text="-M 'msa'"/>
                <has_text text="-A 'muscle'"/>