comparison datasets_gene.xml @ 21:0f3b3813b6ae draft

planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ncbi_datasets commit c1c3f90e4aa7dc258aa61d98ec4eac0b97eef426

20:7b925f7c50b2

    <description>download gene sequences and metadata</description>
    <macros>
        <import>macros.xml</import>
    </macros>
    <expand macro="bio_tools"/>
    <expand macro="requirements"></expand>
    <expand macro="version_command"/>
    <command><![CDATA[
#import re
@SETUP_CERTIFICATES@
datasets download gene $query.subcommand.download_by

    #end if
#end if

#if $filters.fasta_filter_cond.fasta_filter_select
    #if $filters.fasta_filter_cond.fasta_filter_select == 'text'
        --fasta-filter #echo ",".join(f"'{x}'" for x in $filters.fasta_filter_cond.fasta_filter.split(',') if x)
    #else
        --fasta-filter-file '$filters.fasta_filter_cond.fasta_filter_file'
    #end if
#end if

                    <expand macro="text_or_file" what="Gene Symbol" what_extended="NCBI Gene Symbol" help=""/>
                    <expand macro="ortholog"/>
                    <param argument="--taxon" type="text" value="human" label="Species for gene symbol" help="NCBI taxid, common or scientific name">
                        <sanitizer invalid_char="">
                            <valid initial="string.letters">
                                <add value=" " />
                                <add value="-" />
                            </valid>
                        </sanitizer>
                    </param>
                </when>
                <when value="accession">
                    <expand macro="text_or_file" what="Gene Accession" what_extended="NCBI Gene Accession" help=""/>
                    <expand macro="ortholog"/>
                    <param argument="--taxon-filter" type="text" value="" label="Limit gene sequences and annotation report file to specified taxon" help="any rank, only available for WP accessions">
                        <sanitizer invalid_char="">
                            <valid initial="string.letters">
                                <add value=" " />
                                <add value="-" />
                            </valid>
                        </sanitizer>
                    </param>
                    <param argument="--include-flanks-bp" type="integer" optional="true" min="0" value="" label="Length of flanking nucleotides" help="WP accessions only"/>
                </when>

                <when value=""/>
                <when value="text">
                    <param argument="--fasta-filter" type="text" label="RefSeq nucleotide and protein accessions" help="Comma separated">
                        <sanitizer invalid_char="">
                            <valid initial="string.letters,string.digits">
                                <add value="," />
                            </valid>
                        </sanitizer>
                    </param>
                </when>
                <when value="file">

                    <has_n_columns n="8"/>
                </assert_contents>
            </output>
            <output name="rna_fasta">
                <assert_contents>
                    <has_text text=">"/>
                </assert_contents>
            </output>
            <output name="protein_fasta">
                <assert_contents>
                    <has_text text=">"/>
                </assert_contents>
            </output>
        </test>
        <!-- 2: datasets download gene gene-id 2597 14433 -->
        <test expect_num_outputs="3">

                    <has_n_columns n="8"/>
                </assert_contents>
            </output>
            <output name="rna_fasta">
                <assert_contents>
                    <has_text text=">"/>
                </assert_contents>
            </output>
            <output name="protein_fasta">
                <assert_contents>
                    <has_text text=">"/>
                </assert_contents>
            </output>
        </test>
        <!-- 3: same as above + give accessions by file, 2 different outputs and ortholog-->
        <test expect_num_outputs="3">

                </conditional>
                <param name="ortholog" value="Haplorrhini,Strepsirrhini"/>
            </conditional>
            <section name="file_choices">
                <conditional name="kingdom_cond">
                    <param name="include" value="gene,cds"/>
                </conditional>
            </section>
            <output name="gene_data_report">
                <assert_contents>

                    <has_n_columns n="8"/>
                </assert_contents>
            </output>
            <output name="gene_fasta">
                <assert_contents>
                    <has_text text=">"/>
                </assert_contents>
            </output>
            <output name="cds_fasta">
                <assert_contents>
                    <has_text text=">"/>
                </assert_contents>
            </output>
        </test>
        <!-- 4: datasets download gene symbol tp53 -->
        <test expect_num_outputs="1">
            <conditional name="query|subcommand">
                <param name="download_by" value="symbol"/>
                <conditional name="text_or_file">
                    <param name="text_or_file" value="text"/>
                    <param name="accession" value="tp53"/>
                </conditional>
            </conditional>
            <section name="file_choices">
                <conditional name="kingdom_cond">
                    <param name="include" value=""/>
                </conditional>
            </section>
            <output name="gene_data_report">
                <assert_contents>
                    <has_text text="human"/>
                    <has_n_lines n="2"/>
                    <has_n_columns n="8"/>

                    <has_n_columns n="39"/>
                </assert_contents>
            </output>
            <output name="threep_utr_fasta">
                <assert_contents>
                    <has_text text=">"/>
                </assert_contents>
            </output>
            <output name="fivep_utr_fasta">
                <assert_contents>
                    <has_text text=">"/>
                </assert_contents>
            </output>
        </test>
        <!-- 6: datasets download gene symbol brca1 \-\-ortholog -->
        <test expect_num_outputs="1">
            <conditional name="query|subcommand">
                <param name="download_by" value="symbol"/>
                <conditional name="text_or_file">
                    <param name="text_or_file" value="text"/>
                    <param name="accession" value="brca1"/>
                </conditional>
                <param name="ortholog" value="rodentia"/>
            </conditional>
            <section name="file_choices">
                <conditional name="kingdom_cond">
                    <param name="include" value=""/>
                </conditional>
            </section>
            <output name="gene_data_report">
                <assert_contents>
                    <has_text text="rat"/>
                    <has_text text="Brca1"/>
                    <has_n_lines min="30"/>
                    <has_n_columns n="8"/>
                </assert_contents>
            </output>
        </test>
        <!-- 7: datasets download gene accession NP_000483.3 -->
        <test expect_num_outputs="1">
            <conditional name="query|subcommand">
                <param name="download_by" value="accession"/>
                <conditional name="text_or_file">
                    <param name="text_or_file" value="text"/>
                    <param name="accession" value="NP_000483.3"/>
                </conditional>
            </conditional>
            <section name="file_choices">
                <conditional name="kingdom_cond">
                    <param name="include" value=""/>
                </conditional>
            </section>
            <output name="gene_data_report">
                <assert_contents>
                    <has_text text="human"/>
                    <has_n_lines n="2"/>
                    <has_n_columns n="8"/>
                </assert_contents>
            </output>
        </test>
        <!-- 8: datasets download gene accession NM_000546.6 NM_000492.4 + ortholog-->
        <test expect_num_outputs="1">
            <conditional name="query|subcommand">
                <param name="download_by" value="accession"/>
                <conditional name="text_or_file">
                    <param name="text_or_file" value="text"/>
                    <param name="accession" value="NM_000546.6 NM_000492.4"/>
                </conditional>
                <param name="ortholog" value="all"/>
            </conditional>
            <section name="file_choices">
                <conditional name="kingdom_cond">
                    <param name="include" value=""/>
                </conditional>
            </section>
            <output name="gene_data_report">
                <assert_contents>
                    <has_text text="human"/>
                    <has_n_lines min="800"/>
                    <has_n_columns n="8"/>
                </assert_contents>
            </output>
        </test>

        <!-- 9: datasets download gene accession WP_003249567.1 + include_flanks_bp -->
        <test expect_num_outputs="4">
            <conditional name="query|subcommand">
                <param name="download_by" value="accession"/>
                <conditional name="text_or_file">

                    <has_n_columns n="7"/>
                </assert_contents>
            </output>
            <output name="gene_fasta">
                <assert_contents>
                    <has_text text=">"/>
                </assert_contents>
            </output>
            <output name="gene_flanks">
                <assert_contents>
                    <has_text text=">"/>
                </assert_contents>
            </output>
            <output name="protein_fasta">
                <assert_contents>
                    <has_text text=">"/>
                </assert_contents>
            </output>
            <assert_command>
                <has_text text="include-flanks-bp 100"/>
            </assert_command>
        </test> 

        <!-- 10: datasets download gene taxon human   -->
        <!-- <test expect_num_outputs="1">
            <conditional name="query|subcommand">
                <param name="download_by" value="taxon"/>
                <param name="taxon_positional" value="human"/>

21:0f3b3813b6ae

    <description>download gene sequences and metadata</description>
    <macros>
        <import>macros.xml</import>
    </macros>
    <expand macro="bio_tools"/>
    <expand macro="requirements"/>
    <expand macro="version_command"/>
    <command><![CDATA[
#import re
@SETUP_CERTIFICATES@
datasets download gene $query.subcommand.download_by

    #end if
#end if

#if $filters.fasta_filter_cond.fasta_filter_select
    #if $filters.fasta_filter_cond.fasta_filter_select == 'text'
        --fasta-filter #echo ",".join(f"'{x}'" for x in str($filters.fasta_filter_cond.fasta_filter).split(',') if x)
    #else
        --fasta-filter-file '$filters.fasta_filter_cond.fasta_filter_file'
    #end if
#end if

                    <expand macro="text_or_file" what="Gene Symbol" what_extended="NCBI Gene Symbol" help=""/>
                    <expand macro="ortholog"/>
                    <param argument="--taxon" type="text" value="human" label="Species for gene symbol" help="NCBI taxid, common or scientific name">
                        <sanitizer invalid_char="">
                            <valid initial="string.letters">
                                <add value=" "/>
                                <add value="-"/>
                            </valid>
                        </sanitizer>
                    </param>
                </when>
                <when value="accession">
                    <expand macro="text_or_file" what="Gene Accession" what_extended="NCBI Gene Accession" help=""/>
                    <expand macro="ortholog"/>
                    <param argument="--taxon-filter" type="text" value="" label="Limit gene sequences and annotation report file to specified taxon" help="any rank, only available for WP accessions">
                        <sanitizer invalid_char="">
                            <valid initial="string.letters">
                                <add value=" "/>
                                <add value="-"/>
                            </valid>
                        </sanitizer>
                    </param>
                    <param argument="--include-flanks-bp" type="integer" optional="true" min="0" value="" label="Length of flanking nucleotides" help="WP accessions only"/>
                </when>

                <when value=""/>
                <when value="text">
                    <param argument="--fasta-filter" type="text" label="RefSeq nucleotide and protein accessions" help="Comma separated">
                        <sanitizer invalid_char="">
                            <valid initial="string.letters,string.digits">
                                <add value=","/>
                            </valid>
                        </sanitizer>
                    </param>
                </when>
                <when value="file">

                    <has_n_columns n="8"/>
                </assert_contents>
            </output>
            <output name="rna_fasta">
                <assert_contents>
                    <has_text text="&gt;"/>
                </assert_contents>
            </output>
            <output name="protein_fasta">
                <assert_contents>
                    <has_text text="&gt;"/>
                </assert_contents>
            </output>
        </test>
        <!-- 2: datasets download gene gene-id 2597 14433 -->
        <test expect_num_outputs="3">

                    <has_n_columns n="8"/>
                </assert_contents>
            </output>
            <output name="rna_fasta">
                <assert_contents>
                    <has_text text="&gt;"/>
                </assert_contents>
            </output>
            <output name="protein_fasta">
                <assert_contents>
                    <has_text text="&gt;"/>
                </assert_contents>
            </output>
        </test>
        <!-- 3: same as above + give accessions by file, 2 different outputs and ortholog-->
        <test expect_num_outputs="3">

                </conditional>
                <param name="ortholog" value="Haplorrhini,Strepsirrhini"/>
            </conditional>
            <section name="file_choices">
                <conditional name="kingdom_cond">
                    <param name="kingdom_sel" value="gene"/>
                    <param name="include" value="gene,cds"/>
                </conditional>
            </section>
            <output name="gene_data_report">
                <assert_contents>

                    <has_n_columns n="8"/>
                </assert_contents>
            </output>
            <output name="gene_fasta">
                <assert_contents>
                    <has_text text="&gt;"/>
                </assert_contents>
            </output>
            <output name="cds_fasta">
                <assert_contents>
                    <has_text text="&gt;"/>
                </assert_contents>
            </output>
        </test>
        <!-- 4: datasets download gene symbol tp53 -->
        <test expect_num_outputs="3">
            <conditional name="query|subcommand">
                <param name="download_by" value="symbol"/>
                <conditional name="text_or_file">
                    <param name="text_or_file" value="text"/>
                    <param name="accession" value="tp53"/>
                </conditional>
            </conditional>
            <output name="gene_data_report">
                <assert_contents>
                    <has_text text="human"/>
                    <has_n_lines n="2"/>
                    <has_n_columns n="8"/>

                    <has_n_columns n="39"/>
                </assert_contents>
            </output>
            <output name="threep_utr_fasta">
                <assert_contents>
                    <has_text text="&gt;"/>
                </assert_contents>
            </output>
            <output name="fivep_utr_fasta">
                <assert_contents>
                    <has_text text="&gt;"/>
                </assert_contents>
            </output>
        </test>
        <!-- 6: datasets download gene symbol brca1 \-\-ortholog -->
        <test expect_num_outputs="3">
            <conditional name="query|subcommand">
                <param name="download_by" value="symbol"/>
                <conditional name="text_or_file">
                    <param name="text_or_file" value="text"/>
                    <param name="accession" value="brca1"/>
                </conditional>
                <param name="ortholog" value="rodentia"/>
            </conditional>
            <output name="gene_data_report">
                <assert_contents>
                    <has_text text="rat"/>
                    <has_text text="Brca1"/>
                    <has_n_lines min="30"/>
                    <has_n_columns n="8"/>
                </assert_contents>
            </output>
        </test>
        <!-- 7: datasets download gene accession NP_000483.3 -->
        <test expect_num_outputs="3">
            <conditional name="query|subcommand">
                <param name="download_by" value="accession"/>
                <conditional name="text_or_file">
                    <param name="text_or_file" value="text"/>
                    <param name="accession" value="NP_000483.3"/>
                </conditional>
            </conditional>
            <output name="gene_data_report">
                <assert_contents>
                    <has_text text="human"/>
                    <has_n_lines n="2"/>
                    <has_n_columns n="8"/>
                </assert_contents>
            </output>
        </test>
        <!-- 8: datasets download gene accession NM_000546.6 NM_000492.4 + ortholog-->
        <test expect_num_outputs="3">
            <conditional name="query|subcommand">
                <param name="download_by" value="accession"/>
                <conditional name="text_or_file">
                    <param name="text_or_file" value="text"/>
                    <param name="accession" value="NM_000546.6 NM_000492.4"/>
                </conditional>
                <param name="ortholog" value="all"/>
            </conditional>
            <output name="gene_data_report">
                <assert_contents>
                    <has_text text="human"/>
                    <has_n_lines min="800"/>
                    <has_n_columns n="8"/>
                </assert_contents>
            </output>
        </test>
        <!-- 9: datasets download gene accession WP_003249567.1 + include_flanks_bp -->
        <test expect_num_outputs="4">
            <conditional name="query|subcommand">
                <param name="download_by" value="accession"/>
                <conditional name="text_or_file">

                    <has_n_columns n="7"/>
                </assert_contents>
            </output>
            <output name="gene_fasta">
                <assert_contents>
                    <has_text text="&gt;"/>
                </assert_contents>
            </output>
            <output name="gene_flanks">
                <assert_contents>
                    <has_text text="&gt;"/>
                </assert_contents>
            </output>
            <output name="protein_fasta">
                <assert_contents>
                    <has_text text="&gt;"/>
                </assert_contents>
            </output>
            <assert_command>
                <has_text text="include-flanks-bp 100"/>
            </assert_command>
        </test>
        <!-- 10: datasets download gene taxon human   -->
        <!-- <test expect_num_outputs="1">
            <conditional name="query|subcommand">
                <param name="download_by" value="taxon"/>
                <param name="taxon_positional" value="human"/>