comparison filter_by_fasta_ids.xml @ 3:3c623e81be77 draft

planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/filter_by_fasta_ids commit 0556e0fe5aa17c84033a75a45baeb3a4c2b5ff76

2:1bd985f14938

<tool id="filter_by_fasta_ids" name="Filter FASTA" version="2.0">
    <description>on the headers and/or the sequences</description>
    <macros>
        <xml name="regexp_macro" token_label="Regular expression pattern">
            <param name="regexp" type="text" value="" label="@LABEL@" help="Use the Python regular expression syntax as specified in https://docs.python.org/3/library/re.html">
                <validator type="empty_field" />

    <command><![CDATA[
python '$__tool_directory__/filter_by_fasta_ids.py'
-i '$input'
#if $header_criteria.header_criteria_select == 'id_list'
    --id_list '$header_criteria.identifiers'
#elif $header_criteria.header_criteria_select == 'regexp'
    --header_regexp '$header_criteria.regexp'
#end if
#if $sequence_criteria.sequence_criteria_select == 'seq_length'
    --min_length $sequence_criteria.min_length

                <option value="regexp">Regular expression on the headers</option>
            </param>
            <when value="" />
            <when value="id_list">
                <param name="identifiers" type="data" format="txt" label="List of IDs to extract sequences for"/>
            </when>
            <when value="regexp">
                <expand macro="regexp_macro" label="Regular expression pattern the header should match" />
            </when>
        </conditional>

            <param name="input" ftype="fasta" value="input.fasta" />
            <param name="header_criteria_select" value="id_list" />
            <param name="identifiers" ftype="txt" value="ids.txt" />
            <param name="dedup" value="True" />
            <output name="output" file="output_dedup.fasta" />
        </test>
        <test expect_num_outputs="2">
            <param name="input" ftype="fasta" value="input.fasta" />
            <param name="header_criteria_select" value="id_list" />
            <param name="identifiers" ftype="txt" value="ids.txt" />

3:3c623e81be77

<tool id="filter_by_fasta_ids" name="Filter FASTA" version="2.1">
    <description>on the headers and/or the sequences</description>
    <macros>
        <xml name="regexp_macro" token_label="Regular expression pattern">
            <param name="regexp" type="text" value="" label="@LABEL@" help="Use the Python regular expression syntax as specified in https://docs.python.org/3/library/re.html">
                <validator type="empty_field" />

    <command><![CDATA[
python '$__tool_directory__/filter_by_fasta_ids.py'
-i '$input'
#if $header_criteria.header_criteria_select == 'id_list'
    --id_list '$header_criteria.identifiers'
    #if $header_criteria.id_regex.find == 'pattern':
        --pattern '$header_criteria.id_regex.pattern'
    #elif $header_criteria.id_regex.find == 'beginning':
        --pattern '$header_criteria.id_regex.pattern'
    #end if
#elif $header_criteria.header_criteria_select == 'regexp'
    --header_regexp '$header_criteria.regexp'
#end if
#if $sequence_criteria.sequence_criteria_select == 'seq_length'
    --min_length $sequence_criteria.min_length

                <option value="regexp">Regular expression on the headers</option>
            </param>
            <when value="" />
            <when value="id_list">
                <param name="identifiers" type="data" format="txt" label="List of IDs to extract sequences for"/>
                
                
                <conditional name="id_regex">
                    <param name="find" type="select" label="Match IDs by">
                        <option value="beginning">Default: ID is expected at the beginning: &gt;ID </option>
                        <help>Default: &gt;ID will use search pattern >([^| ]+) to input ID; Use custom regex to change</help>
                        <option value="pattern">Custom regex pattern</option>
                    </param>
                    <when value="beginning">
                        <param name="pattern" type="hidden" value=">([^| ]+)" label="regex search pattern for ID" >
                            <sanitizer sanitize="False"/>
                            <validator type="regex" message="must include a group that returns an ID">^.*[(](?![?]:).*[)].*$</validator>
                        </param>
                    </when>
                    <when value="pattern">
                        <param name="pattern" type="text" value="" label="regex search pattern for ID">
                            <help>search pattern must contain %s where the ID will be substituted. Use this for Uniprot Acc: >.+?\|(.+?)\|.*$ </help>
                            <sanitizer sanitize="False"/>
                            <validator type="regex" message="must include a group that returns an ID">^.*[(](?![?]:).*[)].*$</validator>
                        </param>
                    </when>
                </conditional>
                
                
            </when>
            <when value="regexp">
                <expand macro="regexp_macro" label="Regular expression pattern the header should match" />
            </when>
        </conditional>

            <param name="input" ftype="fasta" value="input.fasta" />
            <param name="header_criteria_select" value="id_list" />
            <param name="identifiers" ftype="txt" value="ids.txt" />
            <param name="dedup" value="True" />
            <output name="output" file="output_dedup.fasta" />
        </test>
        <test expect_num_outputs="1">
            <param name="input" ftype="fasta" value="input_sp.fasta" />
            <param name="header_criteria_select" value="id_list" />
            <param name="find" value="pattern" />
            <param name="pattern" value=">.+?\|(.+?)\|.*$" />
            <param name="identifiers" ftype="txt" value="ids_sp.txt" />
            <param name="dedup" value="True" />
            <output name="output" file="output_sp_dedup.fasta" />
        </test>
        <test expect_num_outputs="2">
            <param name="input" ftype="fasta" value="input.fasta" />
            <param name="header_criteria_select" value="id_list" />
            <param name="identifiers" ftype="txt" value="ids.txt" />