<tool id="centrifuge_process" name="centrifuge_process" version="0.1.0" python_template_version="3.5">
    <requirements>
        <requirement type="package" version="3.9">python</requirement>
        <requirement type="package">biopython</requirement>
        <requirement type="package">pandas</requirement>
    </requirements>
    <command detect_errors="exit_code"><![CDATA[
        $__tool_directory__/process_centrifuge_output.py -o '$input1' -r '$input2' -b '$input3' -t processed-centrifuge-bug-ids.txt
    ]]></command>
    <inputs>
        <param label="Centrifuge Output" type="data" name="input1" format="txt" />
	<param label="Centrifuge Report (from --report-file)" type="data" name="input2" format="txt" />
	<param label="Bug name to extract read IDs by" type="text" name="input3" format="txt" />
    </inputs>
    <outputs>
        <data name="output" format="txt" from_work_dir="processed-centrifuge-bug-ids.txt" />
    </outputs>
    <tests>
        <test>
            <param name="input1" value="FAL11127.centrifuge.output.txt"/>
	    <param name="input2" value="FAL11127.centrifuge.report.txt"/>
	    <param name="input3" value="Escherichia coli"/>
            <output name="output" file="processed-centrifuge-bug-ids.txt"/>
        </test>
    </tests>
    <help><![CDATA[
usage: process_centrifuge_output.py [-h] [-v] [-i INPUT_FASTQ] -t TAXA_FILTERED_FASTQ_FILE -r CENT_REPORT -o
                                    CENT_OUTPUT -b BUG

This script is part of the centriflaken pipeline: It processes centrifuge output and produces either a filtered FASTQ
or a text file of FASTQ IDs based on the supplied taxa/bug

optional arguments:
  -h, --help            show this help message and exit
  -v                    For more verbose output
  -i INPUT_FASTQ        Path to input FASTQ file (same as input to centrifuge). If not mentioned, a text file of
                        sequence IDs are produced instead of a FASTQ file
  -t TAXA_FILTERED_FASTQ_FILE
                        Path to output FASTQ or output text file filtered by the taxa specified
  -r CENT_REPORT        Path to centrifuge report
  -o CENT_OUTPUT        Path to centrifuge output
  -b BUG                Name or fragment of name of the bug by which reads are extracted
    ]]></help>
    <citations>
        <citation type="bibtex">
@misc{gitlabcpipescentriflaken,
  author = {Konganti, Kranti},
  year = {2022},
  title = {CPIPES},
  publisher = {GitLab},
  journal = {GitLab repository},
  url = {https://cfsan-git.fda.gov/Kranti.Konganti/cpipes},
}</citation>
    </citations>	
</tool>