<tool id="edu.tamu.cpt2.spanin.generate-putative-usp" name="USP Candidates" version="1.0">
    <description>constructs a putative list of potential u-spanin from an input genomic FASTA</description>
    <macros>
        <import>macros.xml</import>
    </macros>
    <requirements>
        <requirement type="package" version="3.9">python</requirement>
        <requirement type="package" version="1.84">biopython</requirement>
        <requirement type="package" version="2024.9.11">regex</requirement>
    </requirements>
    <command detect_errors="aggressive"><![CDATA[
python '$__tool_directory__/generate-putative-usp.py'
'$fa'
--strand '$strand'
--min_size '$min_size'
--max_size '$max_size'
--lipo_min_start '$lipo_min_start'
--lipo_max_start '$lipo_max_start'
--min_lipo_after '$min_lipo_after'
--max_lipo_after '$max_lipo_after'
--tmd_min_start '$tmd_min_start'
--tmd_max_start '$tmd_max_start'
--tmd_min_size '$tmd_min_size'
--tmd_max_size '$tmd_max_size'
--putative_usp_gff '$putative_usp_gff'
--summary_usp_txt '$summary_usp'
--putative_usp '$putative_usp'
--usp_og '$usp_og'
--usp_ob '$usp_ob'
--usp_op '$usp_op'
--usp_on '$usp_on'
--switch '$switch'
    ]]></command>
    <inputs>
        <param label="FASTA" name="fa" type="data" format="fasta"/>
        <param type="select" label="Strand Choice" name="strand">
            <option value="both">both</option>
            <option value="forward">+</option>
            <option value="reverse">-</option>
        </param>
        <param label="u-spanin minimal length" type="integer" name="min_size" value="100"/>
        <param label="u-spanin maximum length" type="integer" name="max_size" value="200"/>
        <param label="Range Selection; default is all; slices are based on NT. Separated by a colon (eg. #:#)" name="switch" type="text" value="all"/>
        <param label="TMD minimal distance from start codon" type="integer" name="tmd_min_start" value="75"/>
        <param label="TMD maximum distance from start codon" type="integer" name="tmd_max_start" value="200"/>
        <param label="TMD minimal size" type="integer" name="tmd_min_size" value="15"/>
        <param label="TMD maximum size" type="integer" name="tmd_max_size" value="25"/>
        <param label="Lipobox minimal distance from start codon" type="integer" name="lipo_min_start" value="10"/>
        <param label="Lipobox maximum distance from start codon" type="integer" name="lipo_max_start" value="30"/>
        <param label="Minimum amount of periplasmic residues after Lipobox" type="integer" name="min_lipo_after" value="60"/>
        <param label="Maximum amount of periplasmic residues after Lipobox" type="integer" name="max_lipo_after" value="160"/>
    </inputs>
    <outputs>
        <data format="fasta" name="usp_on" label="NucSequences.fa" hidden="true"/>
        <data format="fasta" name="usp_op" label="ProtSequences.fa" hidden="true"/>
        <data format="bed" name="usp_ob" label="BED_Output.bed" hidden="true"/>
        <data format="gff3" name="usp_og" label="GFF_Output.gff" hidden="true"/>
        <data format="fasta" name="putative_usp" label="putative_usp.fa"/>
        <data format="txt" name="summary_usp" label="summary_usp.txt"/>
        <data format="gff3" name="putative_usp_gff" label="putative_usp.gff3"/>
    </outputs>
    <tests>
        <test>
            <param name="fa" value="t1.fa"/>
            <param name="min-size" value="100"/>
            <param name="max-size" value="200"/>
            <param name="switch" value="all"/>
            <param name="tmd_min_start" value="75"/>
            <param name="tmd_max_start" value="200"/>
            <param name="tmd_min_size" value="15"/>
            <param name="tmd_max_size" value="25"/>
            <param name="lipo_min_start" value="15"/>
            <param name="lipo_max_start" value="30"/>
            <param name="min_lipo_after" value="60"/>
            <param name="max_lipo_after" value="160"/>
            <param name="select" value="1"/>
            <output name="usp_on" file="_out_usp.fna"/>
            <output name="usp_op" file="_out_usp.fa"/>
            <output name="usp_ob" file="_out_usp.bed"/>
            <output name="usp_og" file="_out_usp.gff3"/>
            <output name="putative_usp" file="_putative_usp.fa"/>
            <output name="summary_usp" file="_summary_usp.txt"/>
            <output name="putative_usp_gff" file="_putative_usp.gff3"/>
        </test>
    </tests>
    <help><![CDATA[

**What it does**
Searches a genome for candidate u-spanins (USPs), a phage protein involved in outer membrane disruption during Gram-negative bacterial host cell lysis. 


**METHODOLOGY**
This tool combines the methods used in the ISP and OSP Candidates tools to search for putative transmembrane domains and lipoboxes. See those tools for additional detail. Within user-set parameters, the following residues are allowed:

Hydrophobic residues allowed in TMD:

    * [FIWLVMYCATGSP]

Lipobox patterns:

    * [ILMFTV][^REKD][GAS]C
    * A W[AGS]C

**INPUT** --> Genomic FASTA
*Note* - Use a single genome fasta. Multiple FASTAs will not work.

**OUTPUT** -->
Protein sequences which passed the above filters are returned as the candidate USPs in three files: a protein FASTA file, a GFF3, and basic summary statistics text file.
.

    ]]></help>
    <citations>
        <citation type="doi">10.1371/journal.pcbi.1008214</citation>
        <citation type="doi">https://dx.doi.org/10.1016/bs.aivir.2018.09.003</citation>
        <citation type="bibtex">
            @unpublished{galaxyTools,
            author = {C. Ross},
            title = {CPT Galaxy Tools},
            year = {2020-},
            note = {https://github.com/tamu-cpt/galaxy-tools/}
            }
        </citation>
    </citations>
</tool>