<tool id="purge_haplotigs_purge" name="Purge Haplotigs Purge" version="@TOOL_VERSION@+galaxy@VERSION_SUFFIX@" >
    <description>Purge contigs</description>
    <xrefs>
        <xref type='bio.tools'>purgehaplotigs</xref>
    </xrefs>
    <macros>
	  <import>macros.xml</import>
    </macros>
    <expand macro="requirements" />
    <expand macro="stdio" />
    <expand macro="version_command" /> 
    <command><![CDATA[
	    purge_haplotigs purge
	    #if $adv_options.align_cov:
        	-align_cov '$adv_options.align_cov'
	    #end if
	    #if $adv_options.max_match:
	        -max_match '$adv_options.max_match'
	    #end if
	    #if $adv_options.wind_min:
	        -wind_min '$adv_options.wind_min'
	    #end if
	    #if $adv_options.wind_nmax:
	        -wind_nmax '$adv_options.wind_nmax'
	    #end if
	    #if $additional.optional_selector == "use_additional_options":
	    	#if $additional.create_dotplot.include_dotplot =="Yes":
	        	-dotplots
	            	-bam '$additional.create_dotplot.aligned_bam'
	    	#end if
	    	#if $additional.with_repeat.has_repeat =="Yes":
	        	-repeats '$additional.with_repeat.repeat'
	    	#end if
	    #end if
            -t \${GALAXY_SLOTS:-4}
	    -g $genome
	    -c $coverage
	    2>&1

	    ]]></command>
    <inputs>
	    <param name="genome" type="data" format="fasta" label="Genome Assembly" help="Fasta file indexed with samtools faidx"/>
	    <param name="coverage" type="data" format="csv" label="Contig Coverage" help="Contig Coverage file"/>
	    <section name="adv_options" title="Advance Parameters" expanded="False">
		    <param argument="-align_cov" type="integer" value="70" label="Cutoff value to identify haplotigs" help="A cutoff value to identify a contig as a haplotigs in percentage: default=70 (-align_cov)"/>
		    <param argument="-max_match" type="integer" value="250" label="Cutoff value to identify repetitive regions" help="A cutoff value to identify repetitive regions: default=250 (-repeats)"/>
                    <param argument="-wind_min" type="integer" value="5000" label="Minimum window size" help="Minimum window size for BED coverage (dotplots): default=5000 (-wind_min)"/>
                    <param argument="-wind_nmax" type="integer" value="200" label="Max windows per contig" help="Max windows per contig for BED coverage plots (dotplots) : default=200 (-wind_nmax)"/>
	    </section>
	    <conditional name="additional">
		  <param name="optional_selector" type="select" label="Additional Parameters">
			  <option value="no">No</option>
			  <option value="use_additional_options">Yes</option>
		  </param>
		  <when value="no"/>
		  <when value="use_additional_options">
		    <conditional name="with_repeat">
                          <param type="select" name="has_repeat" label="Repetitive region file">
			        <option value="No">no repetitive region file</option>
			        <option value="Yes">with repetitive region file</option>
			  </param>
			  <when value="No"/>
			  <when value="Yes">
			       <param name="repeat" argument="-repeats" type="data" format="bed" label="Repetitive region (BED)" help="repetitive regions in BED file format (-repeats)"/>
			  </when>
		    </conditional>
		    <conditional name="create_dotplot">
			    <param type="select" name="include_dotplot" label="Generate dotplot" help="repetitive regions in BED file format (-repeats)">
				  <option value="No">No</option>
				  <option value="Yes">Yes</option>
			    </param>
			    <when value="No"/>
			    <when value="Yes">
				<param name="aligned_bam" argument="-bam" type="data" format="bam" label="BAM file" help="Alignment file (BAM) to reference genome, required for generating dotplots (-bam)"/>
			    </when>
		    </conditional>
		 </when>
	    </conditional>
    </inputs>
    <outputs>
	    <data name="curated_haplotigs" format="fasta" label="${tool.name} on ${on_string}: curated haplotigs" from_work_dir="curated.haplotigs.fasta"/>
	    <data name="curated_sequences" format="fasta" label="${tool.name} on ${on_string}: curated sequences" from_work_dir="curated.fasta"/>
	    <data name="curated_artefacts" format="fasta" label="${tool.name} on ${on_string}: curated artefacts" from_work_dir="curated.artefacts.fasta"/>
	    <data name="curated_reassignment" format="tsv" label="${tool.name} on ${on_string}: curated assignment" from_work_dir="curated.reassignments.tsv"/>
	    <data name="curated_log" format="txt" label="${tool.name} on ${on_string}: curated log" from_work_dir="curated.contig_associations.log"/>
	    <collection name="output_pngs" type="list" label="${tool.name} on ${on_string}: dotplot diagram">
	 	  <filter> additional['optional_selector'] == "use_additional_options" and create_dotplot['include_dotplot'] == "Yes"</filter>
               <discover_datasets pattern="__name_and_ext__" ext="png" directory="dotplots_reassigned_contigs" visible="false"/>
           </collection>
    </outputs>

    <tests>
	<test>
              <!-- #1 test with common parameters -->
              <param name="genome" value="contigs.fa" ftype="fasta"/>
              <param name="coverage" value="coverage_stats.csv" ftype="csv"/>
              <param name="align_cov" value="70" />
	      <param name="max_match" value="250" />
	      <param name="wind_min" value="5000" />
	      <param name="wind_nmax" value="200" />
	      <output name="curated_haplotigs" file="curated.haplotigs.fasta" ftype="fasta"/>
	      <output name="curated_sequences" file="curated.fasta" ftype="fasta"/>
              <output name="curated_artefacts" file="curated.artefacts.fasta" ftype="fasta"/>
	      <output name="curated_reassignment" file="curated.reassignments.tsv" ftype="tsv"/>
	      <output name="curated_log" file="curated.contig_associations.log" ftype="txt"/>
	      <conditional name="additional">
		      <param name="optional_selector" value="no"/>
	      </conditional>
        </test>
	<test>
              <!-- #2 test with common parameters -->
              <param name="genome" value="contigs.fa" ftype="fasta"/>
              <param name="coverage" value="coverage_stats.csv" ftype="csv"/>
              <param name="align_cov" value="70" />
              <param name="max_match" value="250" />
              <param name="wind_min" value="5000" />
              <param name="wind_nmax" value="200" />
              <output name="curated_haplotigs" file="curated.haplotigs.fasta" ftype="fasta"/>
              <output name="curated_sequences" file="curated.fasta" ftype="fasta"/>
              <output name="curated_artefacts" file="curated.artefacts.fasta" ftype="fasta"/>
              <output name="curated_reassignment" file="curated.reassignments_w_repeats.tsv" ftype="tsv"/>
              <output name="curated_log" file="curated.contig_associations.log" ftype="txt"/>
              <conditional name="additional">
		      <param name="optional_selector" value="use_additional_options"/>
		      <conditional name="with_repeat">
			      <param name="has_repeat" value="Yes"/>
			      <param name="repeat" value="repeats.bed"/>
		      </conditional>
		      <conditional name="create_dotplot">
			      <param name="include_dotplot" value="Yes"/>
			      <param name="dotplots" value="-dotplots"/>
			      <param name="aligned_bam" value="aligned.bam"/>
		      </conditional>
	      </conditional>
	      <output_collection name="output_pngs" type="list" count="5">
		    <element name="000002F" file="dotplots_reassigned_contigs/000002F.png" ftype="png">
			  <assert_content>
				<has_size value="59559" delta="100"/>
			  </assert_content>
	            </element>
		    <element name="000000F_001" file="dotplots_reassigned_contigs/000000F_001.png" ftype="png">
			  <assert_contents>
				<has_size value="31923" delta="100"/>
			  </assert_contents>
		    </element>
		    <element name="000000F_002" file="dotplots_reassigned_contigs/000000F_002.png" ftype="png">
                           <assert_contents>
                                 <has_size value="42614" delta="100"/>
                           </assert_contents>
		    </element>
		    <element name="000000F_003" file="dotplots_reassigned_contigs/000000F_003.png" ftype="png">
                            <assert_contents>
                                  <has_size value="47498" delta="100"/>
                            </assert_contents>
                    </element>
		    <element name="000000F_004" file="dotplots_reassigned_contigs/000000F_004.png" ftype="png">
                            <assert_contents>
                                  <has_size value="32534" delta="100"/>
                            </assert_contents>
                    </element>
	      </output_collection>
        </test>
    </tests>
    <help><![CDATA[

.. class:: infomark

**What it does**

Running the purging pipeline from the purge_haplotigs tool.
	    
**Input**
	    
- input file (g) : Genome assembly in fasta format
- input file (c) : Contig by contig coverage stats csv file from the previous step.

**Parameters**

*OPTIONAL*

- repeats : BED-format file of repeats to ignore during analysis.
- dotplots : Generate dotplots for manual inspection.
- bam : Samtools-indexed bam file of aligned and sorted reads/subreads to the reference, required for generating dotplots.

*ADVANCED*

- align_cov : Percent cutoff for identifying a contig as a haplotig. DEFAULT = 70
- max_match : Percent cutoff for identifying repetitive contigs. Ignored when using repeat annotations (-repeats). DEFAULT = 250
- I         : Minimap2 indexing, drop minimisers every N bases, DEFAULT = 4G
- wind_min  : Min window size for BED coverage plots (for dotplots). DEFAULT = 5000
- wind_nmax : Max windows per contig for BED coverage plots (for dotplots). DEFAULT = 200

**Output**

- curated.haplotigs.fasta
- curated.contig_associations.log
- curated.fasta
- curated.reassignments.tsv
- curated.artefacts.fasta
 
.. class:: infomark
	    
**References**
]]></help>
     <citations>
	    <citation type="doi">DOI: 10.1186/s12859-018-2485-7</citation>
     </citations>
</tool>