<?xml version="1.0"?>
<tool id="edu.tamu.cpt.blast.relatedness.prot" name="Related Genomes" version="19.1.0.0">
	<description>based on protein blast results</description>
	<macros>
		<import>macros.xml</import>
		<import>cpt-macros.xml</import>
	</macros>
	<expand macro="requirements"/>
	<command detect_errors="aggressive">
$__tool_directory__/relatedness_prot.py
${blastIn.blast}
$__tool_directory__/TaxID_List.txt
--hits $hits
$filter
--protein
#if $blastIn.blastType == "XML":
--xmlMode
#end if
--taxFilter "$taxFilter" 
> $accession_list
</command>
	<inputs>
	<conditional name="blastIn">
		<param name="blastType" type="select" label="Blastn Input Type">
                  <option value="XML" selected="true">Blast XML</option>
     		  <option value="TSV">Blast Tabular</option>
                </param>
		<when value="XML">
			<param label="Blastn Results (Blast XML)" name="blast" type="data" format="blastxml"/>
		</when>
		<when value="TSV">
			<param label="Blastn Results" name="blast" type="data" format="tsv,tabular"/>
		</when>
        </conditional>
                <param label = 'Number of results to return' name="hits" type="integer" size="15" value="5"/>
                <param name="filter" type="boolean" truevalue="" falsevalue="--noFilter" checked="true" label="Automatically filter by phage Taxonomy IDs"/>
                <param name="taxFilter" type="text" label="TaxIDs to filter out of results (Space separated)"/>
        </inputs>
	<outputs>
		<data format="tabular" name="accession_list" label="Top BlastP Hits" />
	</outputs>
	<tests>
		<test>
                        <conditional name="blastIn">
                          <param name="blastType" value="TSV"/>
  			  <param name="blast" value="prot_relate_in.tab"/>
                        </conditional>
			<param name="hits" value="10"/>
			<output name="accession_list" file="prot_relate_out.tab" lines_diff="4"/>
		</test>
	</tests>
	<help>
**What it does**

This tool filters a set of BLASTp results and return the top
related genomes based on number of protein-protein matches. The 
default mode is to only consider phage hits (based on TaxID), 
but this can the toggled off. 

The input must be a tabular file from a BLASTp run with the qseqID, 
sallseqid, salltitles, sallacc, and staxIDs fields selected.
		
The output will be a tabular file with the top hits returned, 
where each row specifies the TaxID, organism name, and number of 
similar unique protein hits shared.
</help>
	<expand macro="citations-2020" />
</tool>