--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/parse_mito_blast.xml	Thu May 12 10:45:45 2022 +0000
@@ -0,0 +1,50 @@
+ <tool id="parse_mito_blast" name="Parse mitochondrial blast" version="@TOOL_VERSION@" profile="21.05">
+    <description>overlap-conscious coverage calculation between scaffold-accession number pairs from blast output</description>
+    <macros>
+        <token name="@TOOL_VERSION@">1.0.1</token>
+    </macros>
+    <requirements>
+        <requirement type="package" version="@TOOL_VERSION@">parse_mito_blast</requirement>
+    </requirements>
+    <command detect_errors="exit_code"><![CDATA[
+parse_mito_blast.py
+    --blastout '$blastout' 
+]]>
+    </command>
+    <inputs>
+        <param argument="blastout" type="data" format="tabular" label="Tabular file generated by mito-blast"/>
+    </inputs>
+    <outputs>
+        <data name="cov_report" format="tabular" from_work_dir="cov_report.tsv" label="${tool.name} on ${on_string}" />
+        <data name="mito_scaff_names" format="txt" from_work_dir="mito_scaff_names.txt" label="contaminant scaffolds on ${on_string}" />
+    </outputs>
+    <tests>
+        <test>
+            <param name="blastout" value="test_input.tsv" />
+            <output name="cov_report" file="test_cov_report.tsv" ftype="tabular" /> 
+            <output name="mito_scaff_names" file="test_mito_scaff_names.txt" ftype="txt" />
+        </test>
+    </tests>
+    <help><![CDATA[
+	
+	This tool is intended for use on the tabular (format 6) output generated by blastn. Depending on filtering choices, blast will output \
+	all the alignments it generates. Often there are multiple overlapping alignments for a unique query sequence-accession number pair. This tool \
+	collects all alignments between a given pair and calculates the total coverage between the query and accession number, taking into /
+	account overlapping alignments. Therefore a user can figure out how much of their query sequence actually identifies with a subject /
+	sequence (acc #). This tool was designed for use in the VGP decontamination pipeline - in particular for deciding whether a scaffold /
+	(sequence) is of mitochondrial origin. Parse_mito_blast generates a synthesized, tabular report of coverage for scaffolds with >95% /
+	identity to a mitochondrial subject sequence. To use, pass a format 6 blast output (--blastout) and the desired name of the report. 
+
+	NOTE: This tool is currently designed for use with the blast headers 'qseqid','sseqid','qlen','length','qcovhsp','eval','qstart','qend','qcovs' /
+	and will be sensitive to this order. 
+    ]]>
+    </help>
+    <citations>
+	    <citation type="bibtex">
+           @misc{Danecek_et_al,
+           Author={Danecek, P., Schiffels, S., Durbin, R.},
+           title={Multiallelic calling model in bcftools (-m)},
+           url = {http://samtools.github.io/bcftools/call-m.pdf},}
+        </citation>
+    </citations>
+</tool>