Mercurial > repos > bzeitouni > svdetect

view svdetect/SVDetect_run_parallel.xml @ 22:eb65c0fa5542 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Uploaded
author bzeitouni
date Thu, 12 Jul 2012 12:19:40 -0400
parents 42353a10899a
children
line wrap: on
line source

<tool id="svdetect_run_parallel" name="Detect clusters of anomalously mapped pairs">

<description>and identify structural variants</description>

<command interpreter="perl">SVDetect_run_parallel.pl

#if $getLinks.linking == "linking"
linking
<!-- -out1 '$links_file' -->
#end if
#if $getFilteredLinks.filtering == "filtering"
filtering
<!--- out2 '$flinks_file' -->
#if str($getFilteredLinks.links2SV) == "create"
links2SV
-out3 '$sv_file'
#end if
#if  $getFilteredLinks.file_conversion.file_conversion_select=="convert" and str($getFilteredLinks.file_conversion.links2circos) == "create"
links2circos
-out4 '$circos_file'
#end if
#if  $getFilteredLinks.file_conversion.file_conversion_select=="convert" and str($getFilteredLinks.file_conversion.links2bed) == "create"
links2bed
-out5 '$bed_file'
#end if
#end if
-conf '$config_file'
-l '$log_file'
-N '$sample_name'

</command>

<inputs>
	<param name="sample_name" type="text" value="sample" label="Sample Name"/>
	<param name="mates_file" format="bam" type="data" label="Input BAM file (.ab.bam)"/>
   	<param name="cmap_file" format="len" type="data" label="Chromosomes list file (.len)" help="Tabulated file format with Chromosome ID (integer from 1), name and length"/>
  	<param name="mates_orientation" type="select" format="txt" label="Type of sequencing technology and libraries">
		<option value="FR">Illumina paired-ends</option>
		<option value="RF">Illumina mate-pairs</option>
		<option value="FR">SOLiD paired-ends</option>
		<option value="RR">SOLiD mate-pairs</option>
   	</param>
	<param name="read1_length" type="integer" size="10" value="50" label="Read 1 length (bp)" help="Length of the first read in a pair (left read)"/>
	<param name="read2_length" type="integer" size="10" value="50" label="Read 2 length (bp)" help="Length of the second read in a pair (right read)"/>
	<param name="sv_type" type="select" format="txt" label="Type of SV to detect">
		<option value="all">all types of SVs</option>
		<option value="intra">intrachromosomal SVs only</option>
		<option value="inter">interchromosomal SVs only</option>
  	</param>
   	
   	<conditional name="getLinks">
   		<param name="linking" type="select" label="Linking procedure" help="Detection and isolation of links">
			<option value="linking">Yes</option>
			<option value="">No, already done</option>
  		</param>
		<when value="">
    			<!-- do nothing here -->
   		</when>
    		<when value="linking">
			<param name="splitmate" label="Do you want to split the original mate file per chromosome for parallel computing?" type="boolean" truevalue="split" falsevalue="do_not_split" checked="True" help="Untick it if already done"/>
    			<param name="window_size" type="integer" size="20" value="3000" label="Window size (bp)" help="Equal to at least “2µ+2√2σ"/>
			<param name="step_length" type="integer" size="20" value="250" label="Step length size (bp)" help="Equal to 1/2 or 1/4 of the window size"/>
    		</when>
   	</conditional>

   	<conditional name="getFilteredLinks">
	   	<param name="filtering" type="select" label="Filtering procedure" help="Filtering of links according different parameters and thresholds">
			<option value="filtering">Yes</option>
                        <option value="">No</option>
	  	</param>
		<when value="">
	    		<!-- do nothing here -->
	   	</when>
	    	<when value="filtering">
			
			<param name="splitlink" label="Do you want to split the original link file per chromosome for parallel computing?" type="boolean" truevalue="split" falsevalue="do_not_split" checked="False" help="Untick it if (the linking is) already done"/>
			<param name="chromosomes" type="text" size="20" label="List of chromosome names to keep or exclude"/>
			<param name="nb_pairs_threshold" type="integer" size="20" value="5" label="Minimum number of pairs in a cluster"/>
		
			<conditional name="filter1">
	   			<param name="strand_filtering" type="select" label="Strand filtering procedure">
					<option value="strand">Yes</option>
					<option value="">No</option>
	  			</param>
				<when value="">
	    				<!-- do nothing here -->
	   			</when>
	    			<when value="strand">

					<conditional name="filter2">
			   			<param name="order_filtering" type="select" label="Order filtering procedure">
							<option value="order">Yes</option>
							<option value="">No</option>
			  			</param>
						<when value="">
			    				<!-- do nothing here -->
			   			</when>
			    			<when value="order">

							<conditional name="filter3">
					   			<param name="insert_size_filtering" type="select" label="Insert-size filtering procedure">
									<option value="insert">Yes</option>
									<option value="">No</option>
					  			</param>
								<when value="">
					    				<!-- do nothing here -->
					   			</when>
					    			<when value="insert">
									<param name="indel_sigma_threshold" type="float" size="20" value="3" label="Minimal number of sigma fold for the insert size filtering and to call insertions and deletions"/>
									<param name="dup_sigma_threshold" type="float" size="20" value="3" label="minimal number of sigma fold for the insert size filtering to call tandem duplications"/>
									<param name="singleton_sigma_threshold" type="float" size="20" value="4" label="Minimal number of sigma fold for the insert size filtering to call singletons" help="for Illumina mate-pairs only"/>
			    					</when>
			   				</conditional>

							<param name="mu_length" type="integer" size="20" value="3000" label="Mean insert size value (µ) of normally mapped mate-pairs, in bp"/>
							<param name="sigma_length" type="integer" size="20" value="250" label="Calculated sd value (σ) from the distribution of normally mapped  mate-pairs, in bp"/>
			    				<param name="nb_pairs_order_threshold" type="integer" size="20" value="2" label="Minimal number of pairs in a subgroup of paired-end reads for balanced events"/>
			    			</when>
			   		</conditional>
						
					<param name="final_score_threshold" type="float" size="20"  value="1.0" label="Minimal final filtering score for calling SVs" help="A value of 1 means all the pairs in a cluster were consistent between each other after applying filters"/>
	    			</when>
	   		</conditional>
		
			<param name="links2SV" label="Do you want to have filtered links in a tabulated file format showing significant SVs?" type="boolean" truevalue="create" falsevalue="do_not_create" checked="True"/>
		
			<conditional name="file_conversion">
				<param name="file_conversion_select" type="select" label="Output file conversion" help="Converts filtered links to Circos/BED files format for graphical view of SVs">
					<option value="do_not_convert">No</option>
					<option value="convert">Yes</option>
				</param>
				<when value="do_not_convert">
					    <!-- do nothing here -->
				</when>
				<when value="convert">
					<param name="links2circos" label="Converts the link list to the Circos link format" type="boolean" truevalue="create" falsevalue="do_not_create" checked="True"/>
					<param name="links2bed" label="Converts the link list to the UCSC BED format" type="boolean" truevalue="create" falsevalue="do_not_create" checked="False"/>
					<param name="organism_id" type="text" size="10" value="hs" label="Organism ID"/>
					<repeat name="color_code" title="Color-code" min="1" max="7">
						<param name="color" type="select" label="Color">
							<option value="grey">grey</option>
							<option value="black">black</option>
							<option value="blue">blue</option>
							<option value="green">green</option>
							<option value="purple">purple</option>
							<option value="orange">orange</option>
							<option value="red">red</option>
						</param>
						<param name="interval" type="text" value="1,3" label="Interval"/>
					</repeat>
				</when>
			</conditional>
    		</when>
	</conditional>
</inputs>


<outputs>
	<!--<data format="txt" name="links_file" label="svdetect.links">
		<filter>getLinks['linking']=="linking"</filter>
	</data>
	<data format="txt" name="flinks_file" label="svdetect.links.filtered">
		<filter>getFilteredLinks['filtering']=="filtering"</filter>
	</data>-->
	<data format="sv" name="sv_file" label="${sample_name}.sv">
		<filter>(
			getFilteredLinks['filtering']=="filtering" and
			getFilteredLinks['links2SV'] is True
			)
	         </filter>
	</data>
	<data format="segdup" name="circos_file" label="${sample_name}.segdup">
		<filter>(
			getFilteredLinks['filtering']=="filtering" and
			getFilteredLinks['file_conversion']['file_conversion_select']=="convert" and
			getFilteredLinks['file_conversion']['links2circos'] is True
			)
	         </filter>
	</data>
	<data format="bed" name="bed_file" label="${sample_name}.bed">
		<filter>(
			getFilteredLinks['filtering']=="filtering" and
			getFilteredLinks['file_conversion']['file_conversion_select']=="convert" and
			getFilteredLinks['file_conversion']['links2bed'] is True
			)
	         </filter>
	</data>
	<data format="txt" name="log_file" label="${sample_name}.svdetect_run.log"/>
</outputs>



<configfiles>
	<configfile name="config_file">
&lt;general&gt;
input_format = bam
sv_type = ${sv_type}
mates_orientation=${mates_orientation}
read1_length=${read1_length}
read2_length=${read2_length}
mates_file=${mates_file}
cmap_file=${cmap_file}
tmp_dir=$__new_file_path__/svdetect/tmp
output_dir=$__new_file_path__/svdetect
num_threads=8
&lt;/general&gt; 

#if $getLinks.linking == "linking"
&lt;detection&gt;
#if str($getLinks.splitmate) == "split"
split_mate_file=1
#else
split_mate_file=0
#end if
window_size=${getLinks.window_size}
step_length=${getLinks.step_length}
&lt;/detection&gt; 
#end if

#if $getFilteredLinks.filtering == "filtering"
&lt;filtering&gt;
#if str($getFilteredLinks.splitlink) == "split"
split_link_file=1
#else
split_link_file=0
#end if
#if str($getFilteredLinks.chromosomes) != ""
chromosomes=${getFilteredLinks.chromosomes}
#end if
nb_pairs_threshold=${getFilteredLinks.nb_pairs_threshold}
#if $getFilteredLinks.filter1.strand_filtering == "strand"
strand_filtering=1
final_score_threshold=${getFilteredLinks.filter1.final_score_threshold}
#if $getFilteredLinks.filter1.filter2.order_filtering == "order"
order_filtering=1
mu_length=${getFilteredLinks.filter1.filter2.mu_length}
sigma_length=${getFilteredLinks.filter1.filter2.sigma_length}
nb_pairs_order_threshold=${getFilteredLinks.filter1.filter2.nb_pairs_order_threshold}
#if $getFilteredLinks.filter1.filter2.filter3.insert_size_filtering == "insert"
insert_size_filtering=1
indel_sigma_threshold=${getFilteredLinks.filter1.filter2.filter3.indel_sigma_threshold}
dup_sigma_threshold=${getFilteredLinks.filter1.filter2.filter3.dup_sigma_threshold}
singleton_sigma_threshold=${getFilteredLinks.filter1.filter2.filter3.singleton_sigma_threshold}
#else
insert_size_filtering=0
#end if
#else
order_filtering=0
#end if
#else
strand_filtering=0
#end if
&lt;/filtering&gt; 
#end if

#if $getFilteredLinks.filtering == "filtering"
#if $getFilteredLinks.file_conversion.file_conversion_select == "convert"
#if str($getFilteredLinks.file_conversion.links2circos) == "create"
&lt;circos&gt;
organism_id=${getFilteredLinks.file_conversion.organism_id}
&lt;colorcode&gt;
#for $color_repeat in $getFilteredLinks.file_conversion.color_code
${color_repeat.color}=${color_repeat.interval}
#end for
&lt;/colorcode&gt;
&lt;/circos&gt;
#end if
#if str($getFilteredLinks.file_conversion.links2bed) == "create"
&lt;bed&gt;
&lt;colorcode&gt;
#for $color_repeat in $getFilteredLinks.file_conversion.color_code
#if str($color_repeat.color)== "grey"
190,190,190=${color_repeat.interval}
#end if
#if str($color_repeat.color)== "black"
0,0,0=${color_repeat.interval}
#end if
#if str($color_repeat.color)== "blue"
0,0,255=${color_repeat.interval}
#end if
#if str($color_repeat.color)== "green"
0,255,0=${color_repeat.interval}
#end if
#if str($color_repeat.color)== "purple"
153,50,205=${color_repeat.interval}
#end if
#if str($color_repeat.color)== "orange"
255,140,0=${color_repeat.interval}
#end if
#if str($color_repeat.color)== "red"
255,0,0=${color_repeat.interval}
#end if
#end for
&lt;/colorcode&gt;
&lt;/bed&gt;
#end if
#end if
#end if	
	</configfile>
</configfiles>

  <help>
**What it does**

SVDetect - Version : 0.8

Parallel version (nCPU=8)

SVDetect is a application for the isolation and the type prediction of intra- and inter-chromosomal rearrangements from paired-end/mate-pair sequencing data provided by the high-throughput sequencing technologies

This tool aims to identifying structural variations (SVs) with both clustering and sliding-window strategies, and helping in their visualization at the genome scale.
SVDetect is compatible with SOLiD and Illumina (>=1.3) reads.

Manual documentation available at the http://svdetect.sourceforge.net/Site/Manual.html

-----

.. class:: infomark

Contact Bruno Zeitouni (svdetect@curie.fr) for any questions or concerns about the Galaxy implementation of SVDetect.

  </help>

</tool>