comparison umi-tools_extract.xml @ 14:9fa7803d1c51 draft

"planemo upload commit 59c614a9dda4b9f49bd3349fbde968ee5eca1db6"

13:6a675c3aa610

<tool id="umi_tools_extract" name="UMI-tools extract" version="@VERSION@.1">
    <description>Extract UMI from fastq files</description>
    <macros>
        <import>macros.xml</import>
        <macro name="out_conditional">
            <actions>

                --quality-filter-threshold '$quality.quality_filter_threshold'
                --quality-encoding '$quality.quality_encoding'
            #end if
            #if $print_log == "1":
                --log='$out_log'
            #else
                --supress-stats
            #end if
        #if $gz:
            #if $input_type.type == 'single':
                && mv out.gz '$out'
            #else

        <data name="out_log" format="txt">
            <filter>print_log == True</filter>
        </data>
    </outputs>
    <tests>
        <test>
            <param name="type" value="single" />
            <param name="input_single" value="t_R1.fastq" ftype="fastq" />
            <param name="bc_pattern" value="XXXNNN" />
            <param name="prime3" value="0" />
            <param name="quality_selector" value="true" />

                    <has_text text="umi quality: 28" />
                    <has_text text="Reads output: 72" />
                </assert_contents>
            </output>
        </test>
        <test>
            <param name="type" value="paired" />
            <param name="input_read1" value="t_R1.fastq.gz" ftype="fastq.gz" />
            <param name="input_read2" value="t_R2.fastq.gz" ftype="fastq.gz" />
            <param name="bc_pattern" value="NNNXXX" />
            <output name="out1" file="out_R1.fastq.gz" decompress="true" lines_diff="2" ftype="fastq.gz" />

                    <has_text text="Input Reads: 100" />
                    <has_text text="Reads output: 100" />
                </assert_contents>
            </output>
        </test>
        <test>
            <param name="type" value="paired_collection" /> <!-- same as before, but uncompressed -->
            <param name="paired_type" value="no" />
            <param name="input_readpair" >
                <collection type="paired" >
                    <element name="forward" ftype="fastq" value="t_R1.fastq" />

                    <has_text text="Input Reads: 100" />
                    <has_text text="Reads output: 100" />
                </assert_contents>
            </output>
        </test>
        <test>
            <param name="type" value="paired" />
            <param name="input_read1" value="scrb_seq_fastq.1.gz" ftype="fastq.gz" />
            <param name="input_read2" value="scrb_seq_fastq.2.gz" ftype="fastq.gz" />
            <param name="extract_method" value="string" />
            <param name="bc_pattern" value="CCCCCCNNNNNNNNNN" />
            <param name="use_barcodes" value="yes" />
            <param name="filter_barcode_file" value="scrb_seq_barcodes" />
            <output name="out2" file="scrb_extract.fastq.gz" decompress="true"  ftype="fastq.gz" />
        </test>
        <test><!-- same as above but with regex barcode-->
            <param name="type" value="paired" />
            <param name="input_read1" value="scrb_seq_fastq.1.gz" ftype="fastq.gz" />
            <param name="input_read2" value="scrb_seq_fastq.2.gz" ftype="fastq.gz" />
            <param name="extract_method" value="regex" />
            <param name="bc_pattern" value="^(?P&lt;cell_1&gt;.{6})(?P&lt;umi_1&gt;.{10})" />
            <param name="use_barcodes" value="yes" />
            <param name="filter_barcode_file" value="scrb_seq_barcodes" />
            <output name="out2" file="scrb_extract.fastq.gz" decompress="true" ftype="fastq.gz" />
        </test>
        <test><!-- CelSeq2 example -->
            <param name="type" value="paired" />
            <param name="input_read1" value="read_R1.200.gz" ftype="fastq.gz" />
            <param name="input_read2" value="read_R2.200.gz" ftype="fastq.gz" />
            <param name="extract_method" value="string" />
            <param name="bc_pattern" value="NNNNNNCCCCCC" />
            <output name="out1" file="read_R1.200_extracted.fastq.gz" ftype="fastq.gz" decompress="true" lines_diff="1" />
            <output name="out2" file="read_R2.200_extracted.fastq.gz" ftype="fastq.gz" decompress="true" lines_diff="1" />
        </test>
    </tests>
    <help><![CDATA[

14:9fa7803d1c51

<tool id="umi_tools_extract" name="UMI-tools extract" version="@VERSION@.2">
    <description>Extract UMI from fastq files</description>
    <macros>
        <import>macros.xml</import>
        <macro name="out_conditional">
            <actions>

                --quality-filter-threshold '$quality.quality_filter_threshold'
                --quality-encoding '$quality.quality_encoding'
            #end if
            #if $print_log == "1":
                --log='$out_log'
            #end if
        #if $gz:
            #if $input_type.type == 'single':
                && mv out.gz '$out'
            #else

        <data name="out_log" format="txt">
            <filter>print_log == True</filter>
        </data>
    </outputs>
    <tests>
        <test expect_num_outputs="2">
            <param name="type" value="single" />
            <param name="input_single" value="t_R1.fastq" ftype="fastq" />
            <param name="bc_pattern" value="XXXNNN" />
            <param name="prime3" value="0" />
            <param name="quality_selector" value="true" />

                    <has_text text="umi quality: 28" />
                    <has_text text="Reads output: 72" />
                </assert_contents>
            </output>
        </test>
        <test expect_num_outputs="3">
            <param name="type" value="paired" />
            <param name="input_read1" value="t_R1.fastq.gz" ftype="fastq.gz" />
            <param name="input_read2" value="t_R2.fastq.gz" ftype="fastq.gz" />
            <param name="bc_pattern" value="NNNXXX" />
            <output name="out1" file="out_R1.fastq.gz" decompress="true" lines_diff="2" ftype="fastq.gz" />

                    <has_text text="Input Reads: 100" />
                    <has_text text="Reads output: 100" />
                </assert_contents>
            </output>
        </test>
        <test expect_num_outputs="3">
            <param name="type" value="paired_collection" /> <!-- same as before, but uncompressed -->
            <param name="paired_type" value="no" />
            <param name="input_readpair" >
                <collection type="paired" >
                    <element name="forward" ftype="fastq" value="t_R1.fastq" />

                    <has_text text="Input Reads: 100" />
                    <has_text text="Reads output: 100" />
                </assert_contents>
            </output>
        </test>
        <test expect_num_outputs="3">
            <param name="type" value="paired" />
            <param name="input_read1" value="scrb_seq_fastq.1.gz" ftype="fastq.gz" />
            <param name="input_read2" value="scrb_seq_fastq.2.gz" ftype="fastq.gz" />
            <param name="extract_method" value="string" />
            <param name="bc_pattern" value="CCCCCCNNNNNNNNNN" />
            <param name="use_barcodes" value="yes" />
            <param name="filter_barcode_file" value="scrb_seq_barcodes" />
            <output name="out2" file="scrb_extract.fastq.gz" decompress="true"  ftype="fastq.gz" />
        </test>
        <test expect_num_outputs="3"><!-- same as above but with regex barcode-->
            <param name="type" value="paired" />
            <param name="input_read1" value="scrb_seq_fastq.1.gz" ftype="fastq.gz" />
            <param name="input_read2" value="scrb_seq_fastq.2.gz" ftype="fastq.gz" />
            <param name="extract_method" value="regex" />
            <param name="bc_pattern" value="^(?P&lt;cell_1&gt;.{6})(?P&lt;umi_1&gt;.{10})" />
            <param name="use_barcodes" value="yes" />
            <param name="filter_barcode_file" value="scrb_seq_barcodes" />
            <output name="out2" file="scrb_extract.fastq.gz" decompress="true" ftype="fastq.gz" />
        </test>
        <test expect_num_outputs="2"><!-- CelSeq2 example -->
            <param name="type" value="paired" />
            <param name="input_read1" value="read_R1.200.gz" ftype="fastq.gz" />
            <param name="input_read2" value="read_R2.200.gz" ftype="fastq.gz" />
            <param name="extract_method" value="string" />
            <param name="bc_pattern" value="NNNNNNCCCCCC" />
            <output name="out1" file="read_R1.200_extracted.fastq.gz" ftype="fastq.gz" decompress="true" lines_diff="1" />
            <output name="out2" file="read_R2.200_extracted.fastq.gz" ftype="fastq.gz" decompress="true" lines_diff="1" />
            <param name="print_log" value="false"/>
        </test>
    </tests>
    <help><![CDATA[