comparison fastq_groomer.xml @ 5:4fda2184efa5 draft

"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tool_collections/galaxy_sequence_utils/fastq_groomer commit d4ced60a941c4c4a2fe95de9c09a10086810b387"

4:dafabb4c2037

<tool id="fastq_groomer" name="FASTQ Groomer" version="1.1.1">
    <description>convert between various FASTQ quality formats</description>
    <requirements>
        <requirement type="package" version="1.1.1">galaxy_sequence_utils</requirement>
    </requirements>
    <edam_topics>
        <edam_topic>topic_0622</edam_topic>
    </edam_topics>
    <edam_operations>
        <edam_operation>operation_0233</edam_operation>

#if $input_file.extension.endswith(".gz"):
    #set $suffix = ".gz"
#elif $input_file.extension.endswith(".bz2"):
    #set $suffix = ".bz2"
#else:
#set $suffix = ""
#end if
$input_type$suffix '$output_file'
#if $options_type['options_type_selector'] == 'basic':
    #if str($input_type) == 'cssanger':
        cssanger
    #else:
        sanger
    #end if
    ascii summarize_input
#else:
    ${options_type.output_type} ${options_type.force_quality_encoding} ${options_type.summarize_input}
#end if
    ]]></command>
    <inputs>
        <param name="input_file" type="data" format="fastq,fastq.gz,fastq.bz2" label="File to groom" />
        <param name="input_type" type="select" label="Input FASTQ quality scores type">

                </param>
                <param name="summarize_input" type="select" label="Summarize input data">
                    <option value="summarize_input" selected="true">Summarize Input</option>
                    <option value="dont_summarize_input">Do not Summarize Input (faster)</option>
                </param>
            </when>
        </conditional>
    </inputs>
    <outputs>
        <data name="output_file" format="fastqsanger">

                <when input="options_type.output_type" value="cssanger.bz2" format="fastqcssanger.bz2" />
            </change_format>
        </data>
    </outputs>
    <tests>
        <!-- These tests include test files adapted from supplemental material in Cock PJ, Fields CJ, Goto N, Heuer ML, Rice PM. The Sanger FASTQ file format for sequences with quality scores, and the Solexa/Illumina FASTQ variants. Nucleic Acids Res. 2009 Dec 16. -->
        <!-- Test basic options -->
        <test>
            <param name="input_file" value="sanger_full_range_original_sanger.fastqsanger" ftype="fastq" />
            <param name="input_type" value="sanger" />

    <help><![CDATA[
**What it does**

This tool offers several conversions options relating to the FASTQ format.

When using *Basic* options, the output will be *sanger* formatted or *cssanger* formatted (when the input is Color Space Sanger).

When converting, if a quality score falls outside of the target score range, it will be coerced to the closest available value (i.e. the minimum or maximum).

When converting between Solexa and the other formats, quality scores are mapped between Solexa and PHRED scales using the equations found in `Cock PJ, Fields CJ, Goto N, Heuer ML, Rice PM. The Sanger FASTQ file format for sequences with quality scores, and the Solexa/Illumina FASTQ variants. Nucleic Acids Res. 2009 Dec 16.`_

5:4fda2184efa5

<tool id="fastq_groomer" name="FASTQ Groomer" version="@TOOL_VERSION@">
    <description>convert between various FASTQ quality formats</description>
    <macros>
        <import>macros.xml</import>
    </macros>
    <expand macro="requirements"/>
    <edam_topics>
        <edam_topic>topic_0622</edam_topic>
    </edam_topics>
    <edam_operations>
        <edam_operation>operation_0233</edam_operation>

#if $input_file.extension.endswith(".gz"):
    #set $suffix = ".gz"
#elif $input_file.extension.endswith(".bz2"):
    #set $suffix = ".bz2"
#else:
    #set $suffix = ""
#end if
$input_type$suffix '$output_file'
#if $options_type.options_type_selector == 'basic':
    #if str($input_type) == 'cssanger':
        cssanger
    #else:
        sanger
    #end if
    ascii summarize_input
#else:
    ${options_type.output_type} ${options_type.force_quality_encoding} ${options_type.summarize_input} ${options_type.fix_id}
#end if
    ]]></command>
    <inputs>
        <param name="input_file" type="data" format="fastq,fastq.gz,fastq.bz2" label="File to groom" />
        <param name="input_type" type="select" label="Input FASTQ quality scores type">

                </param>
                <param name="summarize_input" type="select" label="Summarize input data">
                    <option value="summarize_input" selected="true">Summarize Input</option>
                    <option value="dont_summarize_input">Do not Summarize Input (faster)</option>
                </param>
                <param name="fix_id" type="boolean" label="Fix inconsistent identifiers" 
                    truevalue="--fix-id" falsevalue="--no-fix-id" checked="true">
                </param>
            </when>
        </conditional>
    </inputs>
    <outputs>
        <data name="output_file" format="fastqsanger">

                <when input="options_type.output_type" value="cssanger.bz2" format="fastqcssanger.bz2" />
            </change_format>
        </data>
    </outputs>
    <tests>
        <!-- Test fix-id by default -->
        <test>
            <param name="input_file" value="fastq_invalid-line3" ftype="fastq" />
            <param name="input_type" value="sanger" />
            <param name="options_type_selector" value="basic" />
            <output name="output_file" file="fastq_invalid-line3_fixed" />
        </test>
        <!-- Test fix-id by setting the option -->
        <test>
            <param name="input_file" value="fastq_invalid-line3" ftype="fastq" />
            <param name="input_type" value="sanger" />
            <param name="options_type_selector" value="advanced" />
            <param name="fix_id" value="--fix-id" />
            <output name="output_file" file="fastq_invalid-line3_fixed" />
        </test>
        <!-- Test fix-id / option not set; expect failure -->
        <test expect_failure="true">
            <param name="input_file" value="fastq_invalid-line3" ftype="fastq" />
            <param name="input_type" value="sanger" />
            <param name="options_type_selector" value="advanced" />
            <param name="fix_id" value="--no-fix-id" />
        </test>
        <!-- These tests include test files adapted from supplemental material in Cock PJ, Fields CJ, Goto N, Heuer ML, Rice PM. The Sanger FASTQ file format for sequences with quality scores, and the Solexa/Illumina FASTQ variants. Nucleic Acids Res. 2009 Dec 16. -->
        <!-- Test basic options -->
        <test>
            <param name="input_file" value="sanger_full_range_original_sanger.fastqsanger" ftype="fastq" />
            <param name="input_type" value="sanger" />

    <help><![CDATA[
**What it does**

This tool offers several conversions options relating to the FASTQ format.

When using *Basic* options, the output will be *sanger* formatted or *cssanger* formatted (when the input is Color Space Sanger). Inconsistent identifiers are fixed by default.

When converting, if a quality score falls outside of the target score range, it will be coerced to the closest available value (i.e. the minimum or maximum).

When converting between Solexa and the other formats, quality scores are mapped between Solexa and PHRED scales using the equations found in `Cock PJ, Fields CJ, Goto N, Heuer ML, Rice PM. The Sanger FASTQ file format for sequences with quality scores, and the Solexa/Illumina FASTQ variants. Nucleic Acids Res. 2009 Dec 16.`_