--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/generate-putative-isp.xml	Mon Jun 05 02:51:35 2023 +0000
@@ -0,0 +1,81 @@
+<tool id="edu.tamu.cpt2.spanin.generate-putative-isp" name="ISP candidates" version="1.0">
+    <description>constructs a putative list of potential i-spanin from an input genomic FASTA</description>
+    <macros>
+        <import>macros.xml</import>
+        <import>cpt-macros.xml</import>
+    </macros>
+    <expand macro="requirements">
+    </expand>
+    <command detect_errors="aggressive"><![CDATA[
+python '$__tool_directory__/generate-putative-isp.py'
+'$fasta_file'
+--strand '$strand'
+--switch '$switch'
+--isp_on '$isp_on'
+--isp_op '$isp_op'
+--isp_ob '$isp_ob'
+--isp_og '$isp_og'
+--isp_min_len '$isp_min_len'
+--isp_min_dist '$isp_min_dist'
+--isp_max_dist '$isp_max_dist'
+--min_tmd_size '$min_tmd_size'
+--max_tmd_size '$max_tmd_size'
+--putative_isp '$putative_isp'
+--summary_isp_txt '$summary_isp'
+--putative_isp_gff '$putative_isp_gff'
+--isp_max '$isp_max'
+--peri_min '$peri_min'
+--peri_max '$peri_max'
+
+]]></command>
+    <inputs>
+        <param type="select" label="Strand Choice" name="strand">
+            <option value="both">both</option>
+            <option value="forward">+</option>
+            <option value="reverse">-</option>
+        </param>
+        <param label="Single Genome FASTA" name="fasta_file" type="data" format="fasta"/>
+        <param label="i-spanin minimal length" name="isp_min_len" type="integer" value="60"/>
+        <param label="i-spanin maximum length" name="isp_max" type="integer" value="230"/>
+        <param label="Range Selection; default is all; for a specific range to check for a spanin input integers separated by a colon (eg. 1000:2000)" type="text" name="switch" value="all"/>
+        <param label="TMD minimal distance from start codon" name="isp_min_dist" type="integer" value="10"/>
+        <param label="TMD maximum distance from start codon" name="isp_max_dist" type="integer" value="35" help="Searches for a TMD between TMDmin and TMDmax ie [TMDmin,TMDmax]"/>
+        <param label="TMD minimal size" name="min_tmd_size" type="integer" value="10"/>
+        <param label="TMD maximum size" name="max_tmd_size" type="integer" value="25"/>
+        <param label="Periplasmic minimal residue amount" name="peri_min" type="integer" value="16"/>
+        <param label="Periplasmic maximum residue amount" name="peri_max" type="integer" value="206"/>
+    </inputs>
+    <outputs>
+        <data format="fasta" name="isp_on" label="NucSequences.fa" hidden="true"/>
+        <data format="fasta" name="isp_op" label="ProtSequences.fa" hidden="true"/>
+        <data format="bed" name="isp_ob" label="BED_Output.bed" hidden="true"/>
+        <data format="gff3" name="isp_og" label="GFF_Output.gff" hidden="true"/>
+        <data format="fasta" name="putative_isp" label="putative_isp.fa"/>
+        <data format="txt" name="summary_isp" label="summary_isp.txt"/>
+        <data format="gff3" name="putative_isp_gff" label="putative_isp.gff3"/>
+    </outputs>
+    <help><![CDATA[
+
+**What it does**
+Searches a genome for candidate i-spanins (ISPs), a phage protein involved in outer membrane disruption during Gram-negative bacterial host cell lysis.
+
+**METHODOLOGY**
+
+Locates ALL potential start sequences, based on TTG / ATG / GTG (M / L / V). This list is pared down to those within the user-set min/max lengths. That filtered list generates a set of files with the ORFs in FASTA (nt and aa), BED, and GFF3 file formats.
+
+With the protein FASTA, the tool next reads in each potential sequence and determines if it has a putative transmembrane domain (TMD) with the following criteria:
+
+    1. Presence of snorkeling Lysine residues surrounded by hydrophobic residues described for TMD below, within the range the user specifies.
+    2. A putative transmembrane domain, or TMD, defined as a repeated hydrophobic region within the sequence ([FIWLVMYCATGSP]), of length and position within the range the user inputs.
+    3. Length of expected periplasmic region. User defines minimum and maximum thresholds for required number of residues after TMD.
+
+**INPUT** --> Genomic FASTA
+*NOTE: This tool only takes a SINGLE genomic fasta. It does not work with multiFASTAs.*
+
+**OUTPUT** --> putative_isp.fa (FASTA) file, putative_isp.gff3, and basic summary statistics as summary_isp.txt.
+
+Protein sequences which passed the above filters are returned as the candidate ISPs.
+
+]]></help>
+    <expand macro="citations-crr"/>
+</tool>