<tool id="cshl_fastx_clipper" name="Clip" version="1.0.3">
    <description>adapter sequences</description>
    <macros>
        <import>macros.xml</import>
    </macros>
    <expand macro="requirements" />
    <command detect_errors="exit_code"><![CDATA[
@CATS@ fastx_clipper
-l $minlength
-a '$clip_source.clip_sequence'
-d $keepdelta
-o '$output'
-v
$KEEP_N
$DISCARD_OPTIONS
@FQQUAL@
@GZIP@
    ]]></command>

    <inputs>
        <expand macro="fastx_input" />

        <param name="minlength" type="integer" value="15"
            label="Minimum sequence length (after clipping, sequences shorter than this length will be discarded" />

        <conditional name="clip_source">
            <param name="clip_source_list" type="select" label="Source">
                <option value="prebuilt" selected="true">Standard (select from the list below)</option>
                <option value="user">Enter custom sequence</option>
            </param>

            <when value="user">
                <param name="clip_sequence" label="Enter custom clipping sequence" type="text" value="AATTGGCC" />
            </when>

            <when value="prebuilt">
                <param name="clip_sequence" type="select" label="Choose Adapter">
                    <options from_file="fastx_clipper_sequences.txt">
                        <column name="name" index="1"/>
                        <column name="value" index="0"/>
                    </options>
                </param>
            </when>
        </conditional>

        <param name="keepdelta" type="integer" value="0"
            label="Enter a non-zero value to keep the adapter sequence and x bases that follow it"
            help="Use this for hairpin barcoding. keep at 0 unless you know what you're doing" />

        <param name="KEEP_N" type="select" label="Discard sequences with unknown (N) bases">
            <option value="">Yes</option>
            <option value="-n">No</option>
        </param>

        <param name="DISCARD_OPTIONS" type="select" label="Output options">
            <option value="-c">Output only clipped sequences (i.e. sequences which contained the adapter)</option>
            <option value="-C">Output only non-clipped sequences (i.e. sequences which did not contained the adapter)</option>
            <option value="">Output both clipped and non-clipped sequences</option>
        </param>
    </inputs>
    <outputs>
        <data name="output" format_source="input" metadata_source="input" />
    </outputs>
    <tests>
        <test>
            <param name="input" value="fastx_clipper1.fastq" ftype="fastqsolexa"/>
            <param name="maxmismatches" value="2" />
            <param name="minlength" value="15" />
            <param name="clip_source_list" value="user" />
            <param name="clip_sequence" value="CAATTGGTTAATCCCCCTATATA" />
            <param name="keepdelta" value="0" />
            <param name="KEEP_N" value="-n" />
            <param name="DISCARD_OPTIONS" value="-c" />
            <output name="output" ftype="fastqsolexa" file="fastx_clipper1a.out" />
        </test>
    </tests>
    <help><![CDATA[
**What it does**

This tool clips adapters from the 3'-end of the sequences in a FASTA/FASTQ file.

--------


**Clipping Illustration:**

.. image:: fastx_clipper_illustration.png

**Clipping Example:**

.. image:: fastx_clipper_example.png

**In the above example:**

* Sequence no. 1 was discarded since it wasn't clipped (i.e. didn't contain the adapter sequence). (**Output** parameter).
* Sequence no. 5 was discarded --- it's length (after clipping) was shorter than 15 nt (**Minimum Sequence Length** parameter).

------

This tool is based on `FASTX-toolkit`__ by Assaf Gordon.

 .. __: http://hannonlab.cshl.edu/fastx_toolkit/
    ]]></help>
    <expand macro="citations" />
</tool>