File: detect_cnv.xml

package info (click to toggle)
mobyle-programs 5.1.2-1
links: PTS, VCS
area: main
in suites: jessie, jessie-kfreebsd, stretch
size: 5,472 kB
sloc: xml: 126,950; sh: 7; makefile: 4
file content (1254 lines) | stat: -rw-r--r-- 54,993 bytes
parent folder | download | duplicates (3)
<?xml version='1.0' encoding='UTF-8'?>
<!-- XML Authors: Corinne Maufrais                                               -->
<!-- 'Biological Software and Databases' Group, Institut Pasteur, Paris.         -->
<!-- Distributed under LGPLv2 License. Please refer to the COPYING.LIB document. -->
<program xmlns:xi="http://www.w3.org/2001/XInclude">
  <head>
    <name>detect_cnv</name>
    <xi:include href="Entities/penncnv_package.xml"/>
    <doc>
      <title>detect_cnv</title>
      <description>
        <text lang="en">Copy Number Variation (CNV) detection from signal intensity data (Illumina file)</text>
      </description>
    </doc>
    <category>genetics:detection</category>
    <command>kcolumn.pl</command>
  </head>
  <parameters>
    <paragraph>
      <name>preprocess</name>
      <prompt lang="en">Preprocessing of Illumina intensity file for Copy Number Variation (CNV) detection</prompt>
      <parameters>
        <parameter ismandatory="1" issimple="1">
          <name>infile</name>
          <prompt lang="en">Input  signal intensity file</prompt>
          <type>
            <datatype>
              <class>SignalIntensity</class>
              <superclass>AbstractText</superclass>
            </datatype>
          </type>
          <format>
            <code proglang="perl">" $value"</code>
            <code proglang="python">" " + str(value)</code>
          </format>
          <comment>
            <text lang="en">The input signal intensity file is a text file that contains information for one marker per
                         line, and all fields in each line are tab-delimited</text>
            <text lang="en">The first line of the file specifies the meaning for each tab-delimited column. 
        For example, there are six fields in each line in the file, corresponding to SNP name, chromosome, 
        position, genotype, Log R Ratio (LRR) and B Allele Frequency (BAF), respectively.</text>
            <text lang="en">The CNV calling only requires the SNP Name, LRR and BAF values. Note that the relative position
        of LRR and BAF is different from the previous file; again the header line tells the program that the 
        second column represents BAF values, yet the third column is LRR values.</text>
          </comment>
          <argpos>1</argpos>
        </parameter>
        <parameter ismandatory="1" issimple="1">
          <name>split</name>
          <prompt lang="en">Number of tab-delimited column (field) per individual genotyping in intensity file</prompt>
          <type>
            <datatype>
              <class>Integer</class>
            </datatype>
          </type>
          <format>
            <code proglang="perl">" split $value -heading 3 --name_by_header -tab -out $filename"</code>
            <code proglang="python">" split " + str(value) + " -heading 3 --name_by_header -tab --out "+ str(infile)</code>
          </format>
          <argpos>2</argpos>
        </parameter>
      </parameters>
    </paragraph>
    <parameter isout="1" ishidden="1">
      <name>output_split_file</name>
      <prompt lang="en">Split files</prompt>
      <type>
        <datatype>
          <class>Split</class>
          <superclass>AbstractText</superclass>
        </datatype>
      </type>
      <precond>
        <code proglang="perl">defined $infile</code>
        <code proglang="python">infile is not None</code>
      </precond>
      <filenames>
        <code proglang="perl">"$infile.*"</code>
        <code proglang="python">infile + ".*"</code>
      </filenames>
    </parameter>
    <!--  ############### detecnv   -->
    
    
    <paragraph>
      <name>analyse</name>
      <prompt lang="en">Penncnv analyses</prompt>
      <argpos>10</argpos>
      <parameters>
        <parameter issimple="1">
          <name>type</name>
          <prompt lang="en">Analyse type without cnv calls file</prompt>
          <type>
            <datatype>
              <class>Choice</class>
            </datatype>
          </type>
          <vdef>
            <value>null</value>
          </vdef>
          <vlist>
            <velem undef="1">
              <value>null</value>
              <label>Choice</label>
            </velem>
            <velem>
              <value>--test</value>
              <label>Individual-based CNV calling (test)</label>
            </velem>
            <velem>
              <value>--joint</value>
              <label>Joint CNV calls for trio (joint)</label>
            </velem>
            <velem>
              <value>--validate</value>
              <label>Validate copy number at a pre-specified region (validate)</label>
            </velem>
            <velem>
              <value>--summary</value>
              <label>Summary statistics on signal quality (summary)</label>
            </velem>
          </vlist>
          <format>
            <code proglang="perl">(defined $value and $value ne $vdef) ? " &amp;&amp; detect_cnv.pl $value " : ""</code>
            <code proglang="python">( "" , " &amp;&amp; detect_cnv.pl " + str(value) )[ value is not None and value !=vdef]</code>
          </format>
          <ctrl>
            <message>
              <text lang="en">Can not handle both "Analyse type WITH and WITHOUT cnv calls file  at the same time </text>
            </message>
            <code proglang="perl">(defined $value and not defined $rawcnv) or (not defined $value and defined $rawcnv)</code>
            <code proglang="python">(value is not None and rawcnv is None) or (value is None and rawcnv is not None)</code>
          </ctrl>
          <comment>
            <text lang="en">--test: test a signal intensity file to generate CNV calls.</text>
            <text lang="en">--joint: New in July 2008: generate CNV calls for a father-mother-off-
               spring trio via a one-step procedure. It is considerably slower
               than the --trio argument, but generates more accurate CNV calls
               with reduced false negative rates in simulation studies.</text>
            <text lang="en">--summary:   generate summary statistics on signal quality for each input
               file. Usually the summary is provided when calling CNVs and can
               be written to a log file via the --log argument; however, some-
               times users forget to use --log, such that the signal quality
               information is lost. The --summary argument can calculate the
               signal quality again quickly without calling CNVs.</text>
          </comment>
        </parameter>
        <parameter issimple="1">
          <name>rawcnv</name>
          <prompt lang="en">Analyse type with cnv calls file</prompt>
          <type>
            <datatype>
              <class>Choice</class>
            </datatype>
          </type>
          <vdef>
            <value>null</value>
          </vdef>
          <vlist>
            <velem undef="1">
              <value>null</value>
              <label>Choice</label>
            </velem>
            <velem>
              <value>--trio</value>
              <label>Trio-based CNV calling (trio)</label>
            </velem>
            <velem>
              <value>--quartet</value>
              <label>Posterior CNV calls for quartet (quartet)</label>
            </velem>
            <velem>
              <value>--cctest</value>
              <label>Case-control comparison of per-marker CNV frequency (cctest)</label>
            </velem>
            <velem>
              <value>--exclude_heterosomic</value>
              <label>Empirically exclude CNVs in heterosomic chromosomes (exclude_heterosomic)</label>
            </velem>
          </vlist>
          <format>
            <code proglang="perl">(defined $value and $value ne $vdef) ? " &amp;&amp; detect_cnv.pl $value " : ""</code>
            <code proglang="python">( "" , " &amp;&amp; detect_cnv.pl " + str(value) )[ value is not None and value !=vdef]</code>
          </format>
          <ctrl>
            <message>
              <text lang="en">A file containing CNV calls is Mandatory for this parameter</text>
            </message>
            <code proglang="perl">(defined $value and defined $cnvfile and not defined $type) or (not defined $value and defined $type)</code>
            <code proglang="python">(value is not None and cnvfile is not None and type is None) or (value is None and type is not None)</code>
          </ctrl>
          <comment>
            <text lang="en">--trio:  generate CNV calls for a father-mother-offspring trio, given a
               CNV file containing calls generated on each individual separately, a HMM model file, a PFB file, and the three signal
               intensity files.</text>
            <text lang="en">--quartet:    jointly generate CNV calls for a father-mother-offspring1-offp-
               spring2 quartet, given a CNV file containing calls generated on
               each individual separately, a HMM model file, a PFB file, and
               the four signal intensity files.</text>
            <text lang="en"> --cctest: perform a case-control test on the frequency of having CNVs for
               each marker within CNVs. A separate phenotype file must be
               specified via the --phenofile argument for this to work. The
               actual test is a two-sided Fisher exact test. The --onesided
               argument can be specified for performing one-sided test, and
               the --type_filter argument can be specified so that only "dup"
               or "del" is compared between cases and controls.</text>
            <text lang="en">--exclude_heterosomic: exclude CNV calls from chromosomes showing evidence of heterosomic abberations from a given file containing CNV calls. An
               purely empirical method is applied in this procedure, although
               I recommended always manually examine the patterns of BAF to
               determine whether heterosomic abberation is present in a particular sample, if the sample size is relatively small (&lt;100).</text>
          </comment>
          <argpos>10</argpos>
        </parameter>
        <parameter ishidden="1">
          <name>infile_name</name>
          <prompt lang="en">Input file name</prompt>
          <type>
            <datatype>
              <class>Filename</class>
            </datatype>
          </type>
          <precond>
            <code proglang="perl">$type eq '--test' or $type eq '--joint' or $type eq '--validate' or $type eq '--summary' or $rawcnv eq '--trio' or $rawcnv eq '--quartet' or $rawcnv eq '--cctest' or $rawcnv eq '--exclude_heterosomic'</code>
            <code proglang="python">type =='--test' or type =='--joint' or type =='--validate' or type =='--summary' or rawcnv =='--trio' or rawcnv =='--quartet' or rawcnv =='--cctest' or rawcnv =='--exclude_heterosomic'</code>
          </precond>
          <format>
            <code proglang="perl">" $infile.*"</code>
            <code proglang="python">" " + infile + ".*"</code>
          </format>
          <argpos>18</argpos>
          <comment>
            <text lang="en">Specify an output file prefix.</text>
          </comment>
        </parameter>
        
        <parameter issimple="1">
          <name>cnvfile</name>
          <prompt lang="en">CNV calls file (cnv)</prompt>
          <precond>
            <code proglang="perl">defined $rawcnv</code>
            <code proglang="python">rawcnv is not None</code>
          </precond>
          <type>
            <datatype>
              <class>Cnv</class>
              <superclass>AbstractText</superclass>
            </datatype>
          </type>
          <format>
            <code proglang="perl">(defined $value) ? "--cnv $value " : ""</code>
            <code proglang="python">( "" , " --cnv " + str(value) )[ value is not None]</code>
          </format>
          <comment>
            <text lang="en">A file containing CNV calls, that could be generated by the
               -test operation of this program: trio, quartet, exclude_heterosomic, cctest</text>
          </comment>
          <argpos>12</argpos>
        </parameter>
        <parameter ishidden="1">
          <name>out_cnv_filename</name>
          <prompt lang="en">Output CNV file name</prompt>
          <type>
            <datatype>
              <class>Filename</class>
            </datatype>
          </type>
          <precond>
            <code proglang="perl">$type eq '--test' and $infile</code>
            <code proglang="python">type == '--test' and infile</code>
          </precond>
          <format>
            <code proglang="perl">" --out $infile_rawcnv"</code>
            <code proglang="python">" --out " + infile + "_rawcnv "</code>
          </format>
          <argpos>12</argpos>
          <comment>
            <text lang="en">Specify an output file prefix. By default the output filename
               starts with "gengen".</text>
          </comment>
        </parameter>
        <parameter isout="1">
          <name>output_cnv_file</name>
          <prompt lang="en">CNV file</prompt>
          <type>
            <datatype>
              <class>Cnv</class>
              <superclass>AbstractText</superclass>
            </datatype>
          </type>
          <precond>
            <code proglang="perl">$type eq '--test' and $infile</code>
            <code proglang="python">type == '--test' and infile</code>
          </precond>
          <filenames>
            <code proglang="perl">"$infile_rawcnv"</code>
            <code proglang="python">infile + "_rawcnv"</code>
          </filenames>
        </parameter>
        <parameter ismandatory="1" issimple="1">
          <name>hmmmodel</name>
          <prompt lang="en">HMM model (hmm)</prompt>
          <type>
            <datatype>
              <class>Choice</class>
            </datatype>
          </type>
          <precond>
            <code proglang="perl">($type is not None or $rawcnv is not None) and ($type and $rawcnv and $rawcnv ne '--cctest' and $rawcnv ne '--exclude_heterosomic'</code>
            <code proglang="python">(type is not None or rawcnv is not None) and (type != '--summary' and (rawcnv !='--cctest' and rawcnv !='--exclude_heterosomic'))</code>
          </precond>
          <vdef>
            <value>null</value>
          </vdef>
          <vlist>
            <velem undef="1">
              <value>null</value>
              <label>Choose an HMM model</label>
            </velem>
            <velem>
              <value>hhall.hmm</value>
              <label>hhall.hmm</label>
            </velem>
            <velem>
              <value>hh550.hmm</value>
              <label>hh550.hmm</label>
            </velem>
          </vlist>
          <format>
            <code proglang="perl">(defined $value) ? "--hmm <xi:include href="../../Local/Services/Programs/Env/penncnv_data.xml" xpointer="xpointer(/penncnv_data/exec/text())"/>$value " : ""</code>
            <code proglang="python">( "" , " --hmm <xi:include href="../../Local/Services/Programs/Env/penncnv_data.xml" xpointer="xpointer(/penncnv_data/exec/text())"/>" + str(value) )[ value is not None]</code>
          </format>
          <argpos>11</argpos>
          <comment>
            <text lang="en">Specify a HMM model file containing elements necessary for
               specifying the hidden Markov model for CNV calling: test, validate, joint, trio, quartet</text>
          </comment>
        </parameter>
        <parameter ismandatory="1" issimple="1">
          <name>pfb</name>
          <prompt lang="en">Population frequency for B allelel file (pfb)</prompt>
          <type>
            <datatype>
              <class>Choice</class>
            </datatype>
          </type>
          <precond>
            <code proglang="perl">($type is not None or $rawcnv is not None) and ($rawcnv ne '--exclude_heterosomic')</code>
            <code proglang="python">(type is not None or rawcnv is not None) and (rawcnv !='--exclude_heterosomic')</code>
          </precond>
          <vdef>
            <value>null</value>
          </vdef>
          <vlist>
            <velem undef="1">
              <value>null</value>
              <label>Choose an population frequency</label>
            </velem>
            <velem>
              <value>hhall.hg18.pfb</value>
              <label>hhall.hg18.pfb</label>
            </velem>
            <velem>
              <value>hh550.hg18.pfb</value>
              <label>hh550.hg18.pfb</label>
            </velem>
            <velem>
              <value>hc12v1.hg18.pfb</value>
              <label>hc12v1.hg18.pfb</label>
            </velem>
            <velem>
              <value>ho1v1.hg18.pfb</value>
              <label>ho1v1.hg18.pfb</label>
            </velem>
          </vlist>
          <format>
            <code proglang="perl">(defined $value) ? "--pfb <xi:include href="../../Local/Services/Programs/Env/penncnv_data.xml" xpointer="xpointer(/penncnv_data/shared_data/text())"/>$value " : ""</code>
            <code proglang="python">( "" , " --pfb <xi:include href="../../Local/Services/Programs/Env/penncnv_data.xml" xpointer="xpointer(/penncnv_data/shared_data/text())"/>" + str(value) )[ value is not None]</code>
          </format>
          <argpos>11</argpos>
          <comment>
            <text lang="en">A population frequency of B allele file containing chromosome
               coordinates of each SNP, as well as the frequency of B allele
               in a large reference population for this SNP: test, validate, joint, summary, trio, quartet, cctest</text>
          </comment>
        </parameter>
        <parameter issimple="1">
          <name>gcmodel</name>
          <prompt lang="en">A file containing GC model for wave adjustment (gcmodel)</prompt>
          <type>
            <datatype>
              <class>Choice</class>
            </datatype>
          </type>
          <precond>
            <code proglang="perl">$type eq '--test' or $rawcnv eq '--trio' or $rawcnv eq '--quartet' or  $type eq '--joint' or  $type eq '--validate'</code>
            <code proglang="python">(type == '--test' or type =='--joint' or type =='--validate') or (rawcnv =='--trio' or rawcnv =='--quartet')</code>
          </precond>
          <vdef>
            <value>null</value>
          </vdef>
          <vlist>
            <velem undef="1">
              <value>null</value>
              <label>Choose an GC model</label>
            </velem>
            <velem>
              <value>hhall.hg18.gcmodel</value>
              <label>hhall.hg18.gcmodel</label>
            </velem>
            <velem>
              <value>hh550.hg18.gcmodel</value>
              <label>hh550.hg18.gcmodel</label>
            </velem>
            <velem>
              <value>hc12v1.hg18.gcmodel</value>
              <label>hc12v1.hg18.gcmodel</label>
            </velem>
            <velem>
              <value>ho1v1.hg18.gcmodel</value>
              <label>ho1v1.hg18.gcmodel</label>
            </velem>
          </vlist>
          <format>
            <code proglang="perl">(defined $value) ? "--gcmodel <xi:include href="../../Local/Services/Programs/Env/penncnv_data.xml" xpointer="xpointer(/penncnv_data/shared_data/text())"/>$value " : ""</code>
            <code proglang="python">( "" , " --gcmodel <xi:include href="../../Local/Services/Programs/Env/penncnv_data.xml" xpointer="xpointer(/penncnv_data/shared_data/text())"/>" + str(value) )[ value is not None]</code>
          </format>
          <comment>
            <text lang="en">A file that contains the GC percentage in the 1Mb region around
               each marker for the GC-model based signal adjustment: test, joint, validate, trio, quartet</text>
          </comment>
          <argpos>12</argpos>
        </parameter>
      </parameters>
    </paragraph>
    <!--  ############### visualize_cnv   -->
    
    
    
    
    
    
    <paragraph>
      <name>cnvoutput</name>
      <prompt lang="en">CNV output format</prompt>
      <argpos>20</argpos>
      <parameters>
        <parameter>
          <name>outputformat</name>
          <prompt lang="en">Output format</prompt>
          <type>
            <datatype>
              <class>Choice</class>
            </datatype>
          </type>
          <precond>
            <code proglang="perl">($type is not None or $rawcnv is not None)</code>
            <code proglang="python">(type is not None or rawcnv is not None)</code>
          </precond>
          <vdef>
            <value>output</value>
          </vdef>
          <vlist>
            <velem>
              <value>output</value>
              <label>Program output format</label>
            </velem>
            <velem>
              <value>bed</value>
              <label>BED format</label>
            </velem>
            <velem>
              <value>tab</value>
              <label>TAB format</label>
            </velem>
          </vlist>
          <format>
            <code proglang="perl">(defined $value  and $value ne $vdef) ? " &amp;&amp; visualize_cnv.pl --format $value " : ""</code>
            <code proglang="python">( "" , " &amp;&amp; visualize_cnv.pl --format %s " % str(value) )[ value is not None and value !=vdef]</code>
          </format>
        </parameter>
        <parameter isout="1">
          <name>bed_cnv_file</name>
          <prompt lang="en">CNV calls file</prompt>
          <type>
            <datatype>
              <class>Cnv</class>
              <superclass>AbstractText</superclass>
            </datatype>
          </type>
          <precond>
            <code proglang="perl">$type eq '--test' and $infile and $outputformat ne 'output'</code>
            <code proglang="python">type == '--test' and infile and outputformat != 'output'</code>
          </precond>
          <filenames>
            <code proglang="perl">"$infile.$outputformatcnv"</code>
            <code proglang="python">infile + "_" + outputformat + "cnv"</code>
          </filenames>
        </parameter>
        <parameter isout="1">
          <name>bed_file</name>
          <prompt lang="en">Output file</prompt>
          <type>
            <datatype>
              <class>Cnv</class>
              <superclass>AbstractText</superclass>
            </datatype>
          </type>
          <precond>
            <code proglang="perl">($type is not None or $rawcnv is not None) and $type ne '--test' and $outputformat ne 'output'</code>
            <code proglang="python">(type is not None or rawcnv is not None) and type != '--test' and outputformat != 'output'</code>
          </precond>
          <filenames>
            <code proglang="perl">"$infile.$outputformat"</code>
            <code proglang="python">infile + "_" + outputformat</code>
          </filenames>
        </parameter>
        <parameter ishidden="1">
          <name>bed_cnv_infile</name>
          <prompt lang="en">Output file name for visualize_cnv (.rawcnv)</prompt>
          <type>
            <datatype>
              <class>Filename</class>
            </datatype>
          </type>
          <precond>
            <code proglang="perl">$type eq '--test' and $infile and $outputformat ne 'output'</code>
            <code proglang="python">type == '--test' and infile and outputformat != 'output'</code>
          </precond>
          <format>
            <code proglang="perl">" --out $infile.$outputformatcnv $infile.rawcnv "</code>
            <code proglang="python">" --out " + infile + "_" + outputformat + "cnv " + infile + "_rawcnv "</code>
          </format>
          <argpos>21</argpos>
        </parameter>
        <parameter ishidden="1">
          <name>bed_infile</name>
          <prompt lang="en">Output file name for visualize_cnv (.out)</prompt>
          <type>
            <datatype>
              <class>Filename</class>
            </datatype>
          </type>
          <precond>
            <code proglang="perl">($type is not None or $rawcnv is not None) and $type eq '--test' and $outputformat ne 'output'</code>
            <code proglang="python">(type is not None or rawcnv is not None) and type != '--test' and outputformat != 'output'</code>
          </precond>
          <format>
            <code proglang="perl">" --out $infile.$outputformat detect_cnv.out "</code>
            <code proglang="python">" --out " +  infile + "_" + outputformat + " detect_cnv.out " </code>
          </format>
          <argpos>21</argpos>
        </parameter>
      </parameters>
    </paragraph>
    <!-- #################  detect_cnv (suite) -->
    <paragraph>
      <name>cnvcontrol</name>
      <prompt lang="en">CNV output control</prompt>
      <argpos>12</argpos>
      <parameters>
        <parameter>
          <name>minsnp</name>
          <prompt lang="en">Minimum number of SNPs within CNV (minsnp)</prompt>
          <type>
            <datatype>
              <class>Integer</class>
            </datatype>
          </type>
          <precond>
            <code proglang="perl">$type eq '--test' or  $rawcnv eq '--trio' or $rawcnv eq '--quartet' or  $type eq '--joint' or  $type eq '--validate'</code>
            <code proglang="python">type =='--test' or rawcnv =='--trio' or rawcnv =='--quartet' or type =='--joint' or type =='--validate'</code>
          </precond>
          <vdef>
            <value>3</value>
          </vdef>
          <format>
            <code proglang="perl">(defined $value and $value != $vdef) ? "--minsnp $value " : ""</code>
            <code proglang="python">( "" , " --minsnp " + str(value) )[ value is not None and value != vdef]</code>
          </format>
          <comment>
            <text lang="en">The minimum number of SNPs that a CNV call must contain to be
               in output: test, joint, validate, trio, quartet</text>
          </comment>
        </parameter>
        <parameter>
          <name>minlength</name>
          <prompt lang="en">Minimum length of bp within CNV (minlength)</prompt>
          <type>
            <datatype>
              <class>Integer</class>
            </datatype>
          </type>
          <precond>
            <code proglang="perl">$type eq '--test' or $rawcnv eq '--trio' or $rawcnv eq '--quartet' or $type eq '--joint' or  $type eq '--validate'</code>
            <code proglang="python">type =='--test' or rawcnv =='--trio' or rawcnv =='--quartet' or type =='--joint' or type =='--validate'</code>
          </precond>
          <format>
            <code proglang="perl">(defined $value) ? "--minlength $value " : ""</code>
            <code proglang="python">( "" , " --minlength " + str(value) )[ value is not None ]</code>
          </format>
          <ctrl>
            <message>
              <text lang="en">--minlength argument should be a positive integer</text>
            </message>
            <code proglang="perl">$minlength =~ m/^\d+(k|m)?$/i</code>
            <code proglang="python">minlength &gt; 0</code>
          </ctrl>
          <comment>
            <text lang="en">The minimum length of base pairs that a CNV call must contain
               to be in output: test, joint, validate, trio, quartet</text>
          </comment>
        </parameter>
        <parameter>
          <name>minconf</name>
          <prompt lang="en">Minimum confidence score of CNV (minconf)</prompt>
          <type>
            <datatype>
              <class>Integer</class>
            </datatype>
          </type>
          <precond>
            <code proglang="perl">$type eq '--test' or  $type eq '--validate'</code>
            <code proglang="python">type =='--test' or type =='--validate'</code>
          </precond>
          <format>
            <code proglang="perl">(defined $value) ? "--minconf $value " : ""</code>
            <code proglang="python">( "" , " --minconf " + str(value) )[ value is not None ]</code>
          </format>
          <comment>
            <text lang="en">Minimum confidence score for a CNV call to
               be in output. This is an experimental feature, and the
               actual definition of "confidence score" may change in
               the future: test, validate</text>
          </comment>
        </parameter>
        <parameter>
          <name>confidence</name>
          <prompt lang="en">Calculate confidence for each CNV (confidence)</prompt>
          <type>
            <datatype>
              <class>Boolean</class>
            </datatype>
          </type>
          <precond>
            <code proglang="perl">$type eq '--test' or  $type eq '--validate'</code>
            <code proglang="python">type =='--test' or type =='--validate'</code>
          </precond>
          <vdef>
            <value>0</value>
          </vdef>
          <format>
            <code proglang="perl">($value) ? "--confidence" : ""</code>
            <code proglang="python">( "" , " --confidence " )[ value ]</code>
          </format>
          <comment>
            <text lang="en">Calculate a confidence score for each CNV call. This is an
               experimental feature, and the actual definition of "confidence
               score" may change in the future: test, validate</text>
          </comment>
        </parameter>
        <parameter>
          <name>chrx</name>
          <prompt lang="en">Use chromosomeX-specific treatment (chrx)</prompt>
          <type>
            <datatype>
              <class>Boolean</class>
            </datatype>
          </type>
          <precond>
            <code proglang="perl">$type eq '--test' or  $rawcnv eq '--trio' or  $rawcnv eq '--quartet' or  $type eq '--joint' or  $type eq '--validate'</code>
            <code proglang="python">type =='--test' or rawcnv =='--trio' or rawcnv =='--quartet' or type =='--joint' or type =='--validate'</code>
          </precond>
          <vdef>
            <value>0</value>
          </vdef>
          <format>
            <code proglang="perl">($value) ? "--chrx" : ""</code>
            <code proglang="python">( "" , " --chrx " )[ value ]</code>
          </format>
          <comment>
            <text lang="en">Process chromosome X specifically. By default only autosomes
               will be processed by this program: test, joint, validate, trio, quartet.
            </text>
          </comment>
        </parameter>
        <!-- (available soon)
        
        <parameter>  
           <name>chry</name>
           <prompt lang="en">Use chromosomeY-specific treatment</prompt>
           <type>
              <datatype>
                <class>Boolean</class>
              </datatype>
           </type>
           <precond>
        <code proglang="perl">$type eq '-test' or  $rawcnv eq '-trio' or  $rawcnv eq '-quartet' or  $type eq '-joint' or $type eq '-validate'</code>
        <code proglang="python">type =='-test' or rawcnv =='-trio' or rawcnv =='-quartet' or type =='-joint' or type =='-validate'</code>
      </precond>
           <vdef>
            <value>0</value>
           </vdef>
           <format>
               <code proglang="perl">($value) ? "-chry" : ""</code>
               <code proglang="python">( "" , " -chry " )[ value ]</code>
           </format>
           <comment>
            <text lang="en">process chromosome Y specifically. By default only autosomes
               will be processed by this program</text>
               </comment>
        </parameter>
      -->
        <parameter>
          <name>sexfile</name>
          <prompt lang="en">Filename and sex (male/female) for chromosomeX (sex)</prompt>
          <type>
            <datatype>
              <class>Sex</class>
              <superclass>AbstractText</superclass>
            </datatype>
          </type>
          <precond>
            <code proglang="perl">$chrx and not $bafxhet</code>
            <code proglang="python">chrx and not bafxhet</code>
          </precond>
          <format>
            <code proglang="perl">(defined $value) ? "--sex $value " : ""</code>
            <code proglang="python">( "" , " --sex " + str(value) )[ value is not None]</code>
          </format>
          <comment>
            <text lang="en">A 2-column file containing filename and sex (male/female) for
               sex chromosome calling with chromosomeX-specific (chrx) argument. The first tab-
               delimited column should be the input signal file name, while
               the second tab-delimited column should be male or female.
               Alternatively, abbreviations including m (male), f (female), 1
               (male) or 2 (female) are also fine.</text>
          </comment>
          <argpos>12</argpos>
        </parameter>
        <parameter>
          <name>bafxhet</name>
          <prompt lang="en">Minimum BAF heterozygosity rate to predict female gender when file is not supplied (bafxhet)</prompt>
          <type>
            <datatype>
              <class>Float</class>
            </datatype>
          </type>
          <precond>
            <code proglang="perl">$chrx and not $sexfile</code>
            <code proglang="python">chrx and not sexfile</code>
          </precond>
          <vdef>
            <value>0.1</value>
          </vdef>
          <format>
            <code proglang="perl">(defined $value and $value != $vdef) ? "--bafxhet $value" : ""</code>
            <code proglang="python">( "" , " --bafxhet " + str(value) )[ value is not None and value != vdef]</code>
          </format>
          <ctrl>
            <message>
              <text lang="en">Bafxhet argument should be between 0 and 1</text>
            </message>
            <code proglang="perl">$bafxhet_threshold &gt; 0 and $bafxhet_threshold &lt; 1</code>
            <code proglang="python">bafxhet &gt; 0 and bafxhet_threshold &lt; 1</code>
          </ctrl>
          <comment>
            <text lang="en">This argument specifies the BAF heterozygosity rate in chrX to
               predict the sex for a sample. Note that this rate is based on
               BAF values so it is not genotype heterozygosity rate and indeed
               quite different/smaller than that genotype heterozygosity rate.
               By default if &gt;10% chrX markers have BAF values around 0.5, the
               sample is predicted as female. This threshold however does not
               work for Affymetrix genome-wide arrays (instead a 5% threshold
               is better used). For chrX CNV calling, rather than relying on
               PennCNV prediction of gender, it is always best to explicitely
               specify the sample sex using the -sexfile argument.</text>
          </comment>
        </parameter>
      </parameters>
    </paragraph>
    <paragraph>
      <name>validateCalling</name>
      <prompt lang="en">Specific Validation-calling arguments (validate)</prompt>
      <precond>
        <code proglang="perl">$type eq '--validate'</code>
        <code proglang="python">type == '--validate'</code>
      </precond>
      <argpos>12</argpos>
      <parameters>
        <parameter ismandatory="1" issimple="1">
          <name>startsnp</name>
          <prompt lang="en">Start SNP of a pre-specified region (startsnp)</prompt>
          <type>
            <datatype>
              <class>String</class>
            </datatype>
          </type>
          <precond>
            <code proglang="perl">not candlist </code>
            <code proglang="python">not candlist</code>
          </precond>
          <format>
            <code proglang="perl">(defined $value) ? "--startsnp $value" : ""</code>
            <code proglang="python">( "" , " --startsnp " + str(value) )[ value is not None ]</code>
          </format>
          <comment>
            <text lang="en">Specify the start SNP of a pre-specified region used in --validate operation</text>
          </comment>
        </parameter>
        <parameter ismandatory="1" issimple="1">
          <name>endsnp</name>
          <prompt lang="en">End SNP of a pre-specified region (endsnp)</prompt>
          <type>
            <datatype>
              <class>String</class>
            </datatype>
          </type>
          <precond>
            <code proglang="perl">not candlist </code>
            <code proglang="python">not candlist</code>
          </precond>
          <format>
            <code proglang="perl">(defined $value) ? "--endsnp $value" : ""</code>
            <code proglang="python">( "" , " --endsnp " + str(value) )[ value is not None ]</code>
          </format>
          <comment>
            <text lang="en">Specify the end SNP of a pre-specified region used in --validate operation</text>
          </comment>
        </parameter>
        <parameter ismandatory="1" issimple="1">
          <name>delfreq</name>
          <prompt lang="en">Prior deletion frequency of a pre-specified region (delfreq)</prompt>
          <type>
            <datatype>
              <class>Float</class>
            </datatype>
          </type>
          <precond>
            <code proglang="perl">not candlist </code>
            <code proglang="python">not candlist</code>
          </precond>
          <format>
            <code proglang="perl">(defined $value) ? "--delfreq $value" : ""</code>
            <code proglang="python">( "" , " --delfreq " + str(value) )[ value is not None ]</code>
          </format>
          <ctrl>
            <message>
              <text lang="en">Delfreq must be between 0 and 1</text>
            </message>
            <code proglang="perl"> $delfreq &lt; 1 and $delfreq &gt;=0 and (delfrep+dupfreq) &lt;1</code>
            <code proglang="python">delfreq &lt; 1 and delfreq &gt;=0</code>
          </ctrl>
          <comment>
            <text lang="en">Specify the prior deletion allele frequency of a pre-specified
               region used in --validate operation (this frequency can be
               estimated from CNV calls by --test operation)</text>
          </comment>
        </parameter>
        <parameter ismandatory="1" issimple="1">
          <name>dupfreq</name>
          <prompt lang="en">Prior duplication frequency of a pre-specified region (dupfreq)</prompt>
          <type>
            <datatype>
              <class>Float</class>
            </datatype>
          </type>
          <precond>
            <code proglang="perl">not candlist</code>
            <code proglang="python">not candlist</code>
          </precond>
          <format>
            <code proglang="perl">(defined $value) ? "--dupfreq $value" : ""</code>
            <code proglang="python">( "" , " --dupfreq " + str(value) )[ value is not None ]</code>
          </format>
          <ctrl>
            <message>
              <text lang="en">Must be between 0 and 1</text>
            </message>
            <code proglang="perl"> $dupfreq &lt; 1 and $dupfreq &gt;=0</code>
            <code proglang="python">dupfreq &lt; 1 and dupfreq &gt;=0</code>
          </ctrl>
          <comment>
            <text lang="en">Specify the prior duplication allele frequency of a pre-specified region used 
            in --validate operation (this frequency can be estimated from CNV calls by --test operation)</text>
          </comment>
        </parameter>
        <parameter>
          <name>backfreq</name>
          <prompt lang="en">Background CNV probability for any loci (backfreq)</prompt>
          <type>
            <datatype>
              <class>Float</class>
            </datatype>
          </type>
          <precond>
            <code proglang="perl">not $delfreq and not $dupfreq </code>
            <code proglang="python">not delfreq and not dupfreq</code>
          </precond>
          <vdef>
            <value>0.0001</value>
          </vdef>
          <format>
            <code proglang="perl">(defined $value) ? "--backfreq $value" : ""</code>
            <code proglang="python">( "" , " --backfreq " + str(value) )[ value is not None ]</code>
          </format>
          <ctrl>
            <message>
              <text lang="en">--backfreq argument should be less than 0.5</text>
            </message>
            <code proglang="perl">$backfreq &gt; 0 and $backfreq &lt; 0.5</code>
            <code proglang="python">backfreq &gt; 0 and backfreq &lt; 0.5</code>
          </ctrl>
          <comment>
            <text lang="en">Background CNV probability for any loci, with default value as
               0.0001. This argument is useful in validation calling. When
               -delfreq/-dupfreq is not specified, the background frequency is
               used to calculate the prior probability of different copy number states.</text>
          </comment>
        </parameter>
        <parameter>
          <name>candlist</name>
          <prompt lang="en">A file containing all candidate CNV regions to be validated (candlist)</prompt>
          <type>
            <datatype>
              <class>CandidateRegion</class>
              <superclass>AbstractText</superclass>
            </datatype>
          </type>
          <precond>
            <code proglang="perl">not $startsnp and not $endsnp and not $delfreq and not $dupfreq</code>
            <code proglang="python">not startsnp and not endsnp and not delfreq and not dupfreq</code>
          </precond>
          <format>
            <code proglang="perl">(defined $value) ? "--candlist $value " : ""</code>
            <code proglang="python">( "" , " --candlist " + str(value) )[ value is not None]</code>
          </format>
        </parameter>
      </parameters>
    </paragraph>
    <paragraph>
      <name>cctestCalling</name>
      <prompt lang="en">Specific Case-control comparison arguments (cctest)</prompt>
      <precond>
        <code proglang="perl">$rawcnv eq '--cctest'</code>
        <code proglang="python">rawcnv == '--cctest'</code>
      </precond>
      <argpos>12</argpos>
      <parameters>
        <parameter ismandatory="1" issimple="1">
          <name>phenofile</name>
          <prompt lang="en">A file containing phenotype information for each input file (phenofile)</prompt>
          <type>
            <datatype>
              <class>Phenotype</class>
              <superclass>AbstractText</superclass>
            </datatype>
          </type>
          <precond>
            <code proglang="perl">$rawcnv eq '--cctest'</code>
            <code proglang="python">rawcnv == '--cctest'</code>
          </precond>
          <format>
            <code proglang="perl">(defined $value) ? "--phenofile $value " : ""</code>
            <code proglang="python">( "" , " --phenofile " + str(value) )[ value is not None]</code>
          </format>
          <comment>
            <text lang="en">A file containing phenotype informatoin for each individual, so
               that --cctest can be used to compare the frequency between
               cases and controls. Each line has two tab-delimited fields:
               file name and the phenotype. By default, "control" means
               control subjects, and other words means cases; however, the
               user can use --control_label argument to change the phenotype
               label for controls.
        </text>
          </comment>
          <argpos>12</argpos>
        </parameter>
        <parameter>
          <name>control_label</name>
          <prompt lang="en">The phenotype label for control subjects in the phenotype file (control_label)</prompt>
          <type>
            <datatype>
              <class>String</class>
            </datatype>
          </type>
          <format>
            <code proglang="perl">(defined $value) ? "--control_label $value " : ""</code>
            <code proglang="python">( "" , " --control_label " + str(value) )[ value is not None ]</code>
          </format>
          <comment>
            <text lang="en">Specify the text label for control subjects in the phenotype
               file specified by the --phenofile argument. Normally the "control" is used to specify 
               controls, and all other individuals
               are treated as cases. However, some times users may use 1 to
               denote controls and 2 to denote cases; in such situations the
               "--control_label 1" should be used for the --cctest operation.</text>
          </comment>
        </parameter>
        <parameter>
          <name>onesided</name>
          <prompt lang="en">Performed one-sided test (onesided)</prompt>
          <type>
            <datatype>
              <class>Boolean</class>
            </datatype>
          </type>
          <vdef>
            <value>0</value>
          </vdef>
          <format>
            <code proglang="perl">($value) ? "--onesided" : ""</code>
            <code proglang="python">( "" , " --onesided " )[ value ]</code>
          </format>
        </parameter>
        <parameter>
          <name>type_filter</name>
          <prompt lang="en">Used together to specify types of CNVs to be tested (type_filter)</prompt>
          <type>
            <datatype>
              <class>Choice</class>
            </datatype>
          </type>
          <vdef>
            <value>null</value>
          </vdef>
          <vlist>
            <velem undef="1">
              <value>null</value>
              <label>Both duplications and deletions</label>
            </velem>
            <velem>
              <value>dup</value>
              <label>Duplications</label>
            </velem>
            <velem>
              <value>del</value>
              <label>Deletions</label>
            </velem>
          </vlist>
          <format>
            <code proglang="perl">(defined $value  and $value ne $vdef) ? "--type_filter $value " : ""</code>
            <code proglang="python">( "" , " --type_filter " + str(value) )[ value is not None and value !=vdef ]</code>
          </format>
          <comment>
            <text lang="en">Specify the particular types of CNVs to be used in the --cctest
               operation. By default both duplications and deletions are
               treated as a single group of CNVs and be used to compare cases
               and controls.
            </text>
          </comment>
        </parameter>
      </parameters>
    </paragraph>
    <paragraph>
      <name>miscOpt</name>
      <prompt lang="en">Misc options</prompt>
      <argpos>12</argpos>
      <parameters>
        <parameter>
          <name>fmprior</name>
          <prompt lang="en">Prior belief on CN state for regions with CNV calls. Six numbers separated by a comma (fmprior)</prompt>
          <type>
            <datatype>
              <class>String</class>
            </datatype>
          </type>
          <precond>
            <code proglang="perl">$rawcnv eq '--trio' or  $rawcnv eq '--quartet'</code>
            <code proglang="python">rawcnv =='--trio' or rawcnv =='--quartet'</code>
          </precond>
          <format>
            <code proglang="perl">(defined $value) ? "--fmprior $value " : ""</code>
            <code proglang="python">( "" , " --fmprior " + str(value) )[ value is not None ]</code>
          </format>
          <ctrl>
            <message>
              <text lang="en">The --fmprior argument should be 6 comma-separated numbers that sum up to 1.</text>
            </message>
            <code proglang="perl">$value ~= /\d+(,\d+){5}/</code>
            <code proglang="python">len(value.split(',')) == 6 </code>
          </ctrl>
          <comment>
            <text lang="en">The prior probability of 6 hidden states a given CNV call in
               father or mother.  This is used for joint calling of trios or
               quartets. It is specified as six numbers separated by a comma that sum up to 1.
               The empirically derived default values actually work well: trio, quartet.</text>
          </comment>
        </parameter>
        <parameter>
          <name>denovo_rate</name>
          <prompt lang="en">Prior belief on genome-wide de novo event rate (denovo_rate)</prompt>
          <type>
            <datatype>
              <class>Float</class>
            </datatype>
          </type>
          <precond>
            <code proglang="perl">$rawcnv eq '--trio' or  $rawcnv eq '--quartet'</code>
            <code proglang="python">rawcnv =='--trio' or rawcnv =='--quartet'</code>
          </precond>
          <vdef>
            <value>0.0001</value>
          </vdef>
          <format>
            <code proglang="perl">(defined $value  and $value != $vdef) ? "--denovo_rate $value" : ""</code>
            <code proglang="python">( "" , " --denovo_rate " + str(value) )[ value is not None and value != vdef]</code>
          </format>
          <comment>
            <text lang="en">Specify the probability that a given CNV is a de novo event for
               family-based CNV calling. The default is 0.0001. trio, quartet</text>
          </comment>
        </parameter>
        <parameter>
          <name>medianadjust</name>
          <prompt lang="en">Adjust genome-wide LRR such that median equal 0 (nomedianadjust)</prompt>
          <type>
            <datatype>
              <class>Boolean</class>
            </datatype>
          </type>
          <precond>
            <code proglang="perl">$type eq '--test' or $rawcnv eq '--trio' or $rawcnv eq '--quartet' or $type eq '--joint' or  $type eq '--validate'</code>
            <code proglang="python">type =='--test' or rawcnv =='--trio' or rawcnv =='--quartet' or type =='--joint' or type =='--validate'</code>
          </precond>
          <vdef>
            <value>1</value>
          </vdef>
          <format>
            <code proglang="perl">(not $value) ? "--nomedianadjust" : ""</code>
            <code proglang="python">( "" , " --nomedianadjust " )[ not value ]</code>
          </format>
          <comment>
            <text lang="en">This option is turned on by default. It adjust the log R Ratio
               values of the entire genome by a constant so that the median is
               zero: test, trio, quartet, joint, validate.
            </text>
          </comment>
        </parameter>
        <parameter>
          <name>bafadjust</name>
          <prompt lang="en">Adjust genome-wide BAF such that median equal 0.5 (nobafadjust)</prompt>
          <type>
            <datatype>
              <class>Boolean</class>
            </datatype>
          </type>
          <precond>
            <code proglang="perl">$type eq '--test' or $rawcnv eq '--trio' or $rawcnv eq '--quartet' or  $type eq '--joint' or  $type eq '--validate'</code>
            <code proglang="python">type =='--test' or rawcnv =='--trio' or rawcnv =='--quartet' or type =='--joint' or type =='--validate'</code>
          </precond>
          <vdef>
            <value>1</value>
          </vdef>
          <format>
            <code proglang="perl">(not $value) ? "--nobafadjust" : ""</code>
            <code proglang="python">( "" , " --nobafadjust " )[ not value ]</code>
          </format>
          <comment>
            <text lang="en">This option is turned ON by default (new July 2008): it adjust
               the BAF values genome-wide such that the median value is 0.5.</text>
          </comment>
        </parameter>
        <parameter>
          <name>sdadjust</name>
          <prompt lang="en">Adjust SD of hidden Markov model based on input signal (nosdadjust)</prompt>
          <type>
            <datatype>
              <class>Boolean</class>
            </datatype>
          </type>
          <precond>
            <code proglang="perl">$type eq '--test' or $rawcnv eq '--trio' or $rawcnv eq '--quartet' or $type eq '--joint' or  $type eq '--validate'</code>
            <code proglang="python">type =='--test' or rawcnv =='--trio' or rawcnv =='--quartet' or type =='--joint' or type =='--validate'</code>
          </precond>
          <vdef>
            <value>1</value>
          </vdef>
          <format>
            <code proglang="perl">(not $value) ? "--nosdadjust" : ""</code>
            <code proglang="python">( "" , " --nosdadjust " )[ not value ]</code>
          </format>
          <comment>
            <text lang="en">This option is turned ON by default: it adjust the SD values in
               HMM model such that the model fits the signal quality of the
               testing sample to reduce false positive calls</text>
          </comment>
        </parameter>
        <parameter>
          <name>flush</name>
          <prompt lang="en">Flush input/output buffer (noflush)</prompt>
          <type>
            <datatype>
              <class>Boolean</class>
            </datatype>
          </type>
          <vdef>
            <value>1</value>
          </vdef>
          <format>
            <code proglang="perl">(not $value) ? "--noflush" : ""</code>
            <code proglang="python">( "" , " --noflush " )[ not value ]</code>
          </format>
          <comment>
            <text lang="en">This argument is turned ON by default. It requires the
               input/output buffer to flush immediately (that is, no
               input/output is buffered). When PennCNV is running remotely
               (for example, through a SSH connection) or when the output is
               redirected, this argument cause the program to report progress
               in real-time.  When running PennCNV in parallel with many processes 
               accessing disks simultaneously, this option should be
               turned off to decrease system overhead.
            </text>
          </comment>
        </parameter>
      </parameters>
    </paragraph>
    <parameter ishidden="1">
      <name>outfile_name</name>
      <prompt lang="en">Outfile for detect_cnv and Input file for visualize_cnv</prompt>
      <type>
        <datatype>
          <class>Filename</class>
        </datatype>
      </type>
      <precond>
        <code proglang="perl">($type is not None or $rawcnv is not None) and $type ne '--test'</code>
        <code proglang="python">(type is not None or rawcnv is not None) and type != '--test'</code>
      </precond>
      <format>
        <code proglang="perl">" --out detect_cnv.out "</code>
        <code proglang="python">" --out detect_cnv.out "</code>
      </format>
      <argpos>17</argpos>
    </parameter>
    <parameter isstdout="1">
      <name>output_file</name>
      <prompt lang="en">Output file</prompt>
      <type>
        <datatype>
          <class>Cnv</class>
          <superclass>AbstractText</superclass>
        </datatype>
      </type>
      <precond>
        <code proglang="perl">$type ne '--test'</code>
        <code proglang="python">type != '--test'</code>
      </precond>
      <filenames>
        <code proglang="perl">"detect_cnv.out"</code>
        <code proglang="python">"detect_cnv.out"</code>
      </filenames>
    </parameter>
  </parameters>
</program>