/var/lib/mobyle/programs/scan

<?xml version='1.0' encoding='UTF-8'?>
<!-- XML Authors: Corinne Maufrais, Nicolas Joly and Bertrand Neron,             -->
<!-- '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>scan_region</name>
    <xi:include href="Entities/penncnv_package.xml"/>
    <doc>
      <title>scan_region</title>
      <description>
        <text lang="en">Scan genomic regions in a query-file against a DB-file which contains
        chromosome locations for various genomics features</text>
      </description>
    </doc>
    <category>genetics:detection</category>
    <command>scan_region.pl</command>
  </head>
  <parameters>
    <parameter ismandatory="1" issimple="1">
      <name>cnvfile</name>
      <prompt lang="en">CNV calls file (cnv)</prompt>
      <type>
        <datatype>
          <class>Cnv</class>
          <superclass>AbstractText</superclass>
        </datatype>
      </type>
      <format>
        <code proglang="perl">(defined $value) ? " $value " : ""</code>
        <code proglang="python">( "" , " " + 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 detect_cnv program.</text>
      </comment>
      <argpos>1</argpos>
    </parameter>
    <parameter ismandatory="1" issimple="1">
      <name>reffile</name>
      <prompt lang="en">Reference genes for CNV calls generated using hg18 (Mar 2006, NCBI build 36) human genome assembly</prompt>
      <type>
        <datatype>
          <class>Choice</class>
        </datatype>
      </type>
      <vdef>
        <value>null</value>
      </vdef>
      <vlist>
        <velem undef="1">
          <value>null</value>
          <label>Choose a Reference genes databases</label>
        </velem>
        <velem>
          <value>hg18_refGene.txt</value>
          <label>UCSC refGene Annotation</label>
        </velem>
        <velem>
          <value>UCSCknownGene.txt</value>
          <label>UCSC knownGene Annotation</label>
        </velem>
      </vlist>
      <format>
        <code proglang="perl">(defined $value) ? " <xi:include href="../../Local/Services/Programs/Env/penncnv_data.xml" xpointer="xpointer(/penncnv_data/shared_data/text())"/>$value " : ""</code>
        <code proglang="python">( "" , " <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>2</argpos>
    </parameter>
    
    <parameter ismandatory="1" issimple="1">
      <name>reference</name>
      <prompt lang="en">Flags specifying type of databases</prompt>
      <type>
        <datatype>
          <class>Choice</class>
        </datatype>
      </type>
      <vdef>
        <value>null</value>
      </vdef>
      <vlist>
        <velem undef="1">
          <value>null</value>
          <label>Choose a flags specifying type of databases</label>
        </velem>
        <velem>
          <value>--refgene</value>
          <label>Database is UCSC refGene annotation file</label>
        </velem>
        <velem>
          <value>--refcds</value>
          <label>Database is UCSC refGene with exon annotation</label>
        </velem>
        <velem>
          <value>--refexon</value>
          <label>Database is UCSC refGene with coding sequence annotation</label>
        </velem>
        <velem>
          <value>--knowngene</value>
          <label>Database is UCSC knownGene annotation file</label>
        </velem>
      </vlist>
      <format>
        <code proglang="perl">(defined $value and $value ne $vdef) ? " $value " : ""</code>
        <code proglang="python">( "" , " " + str(value) )[ value is not None  and value !=vdef]</code>
      </format>
      <argpos>3</argpos>
      <comment>
        <text lang="en">--refgene:  specify that the database file is in refGene format from UCSC genome
               browser.</text>
        <text lang="en">--refcds:  specify that the database file is in refGene format from UCSC genome
               browser, but user is only interested in the overlap of coding
               region (first exon to last exon).</text>
        <text lang="en">--refexon:   specify that the database file is in refGene format from UCSC genome
               browser, but user is only interested in the overlap of query
               with exons.</text>
        <text lang="en">--knowngene:  specify that the database file is in knownGene format from UCSC
               genome browser.</text>
      </comment>
    </parameter>
    <paragraph>
      <name>dbfile-specific</name>
      <prompt lang="en">Database-specific arguments</prompt>
      <parameters>
        <parameter>
          <name>name2</name>
          <prompt lang="en">Use name2 annotation in refGene file in output</prompt>
          <type>
            <datatype>
              <class>Boolean</class>
            </datatype>
          </type>
          <precond>
            <code proglang="perl">$reference eq '--refgene'</code>
            <code proglang="python">reference == '--refgene'</code>
          </precond>
          <vdef>
            <value>0</value>
          </vdef>
          <format>
            <code proglang="perl">($value) ? "--name2  " : ""</code>
            <code proglang="python">( "" , " --name2  " )[ value]</code>
          </format>
          <argpos>3</argpos>
          <comment>
            <text lang="en">This argument is used in conjunction with the --refgene argument,
         to specify that the alternative gene symbol in the
               "name2" field in the refGene file be printed in the output.</text>
          </comment>
        </parameter>
        <parameter>
          <name>reflink</name>
          <prompt lang="en">Specify a cross-reference file for the RefGene track in UCSC genome browser</prompt>
          <type>
            <datatype>
              <class>Boolean</class>
            </datatype>
          </type>
          <precond>
            <code proglang="perl">$reference ne '--knowngene' </code>
            <code proglang="python">reference != '--knowngene'</code>
          </precond>
          <vdef>
            <value>0</value>
          </vdef>
          <format>
            <code proglang="perl">($value) ? "--reflink <xi:include href="../../Local/Services/Programs/Env/penncnv_data.xml" xpointer="xpointer(/penncnv_data/shared_data/text())"/>hg18_refLink.txt " : ""</code>
            <code proglang="python">( "" , " --reflink <xi:include href="../../Local/Services/Programs/Env/penncnv_data.xml" xpointer="xpointer(/penncnv_data/shared_data/text())"/>hg18_refLink.txt " )[ value]</code>
          </format>
          <argpos>3</argpos>
          <comment>
            <text lang="en">Specify a cross-reference file for the RefGene track in UCSC
               genome browser, so that in the output, the gene identifier
               (gene name or refseq id) are replaced by the gene symbol specified
                in the link file. (If not found in the reflink file, the
               gene identifiers are still used).</text>
          </comment>
        </parameter>
        <parameter>
          <name>kgxref</name>
          <prompt lang="en">Specify a cross-reference file for the knownGene track in UCSC  genome browser</prompt>
          <type>
            <datatype>
              <class>Boolean</class>
            </datatype>
          </type>
          <precond>
            <code proglang="perl">$reference eq '--knowngene'</code>
            <code proglang="python">reference == '--knowngene'</code>
          </precond>
          <vdef>
            <value>0</value>
          </vdef>
          <format>
            <code proglang="perl">($value) ? "--kgxref <xi:include href="../../Local/Services/Programs/Env/penncnv_data.xml" xpointer="xpointer(/penncnv_data/shared_data/text())"/>UCSCkgXref.txt " : ""</code>
            <code proglang="python">( "" , " --kgxref <xi:include href="../../Local/Services/Programs/Env/penncnv_data.xml" xpointer="xpointer(/penncnv_data/shared_data/text())"/>UCSCkgXref.txt " )[ value]</code>
          </format>
          <argpos>3</argpos>
          <comment>
            <text lang="en">Specify a cross-reference file for the knownGene track in UCSC
               genome browser, so that in the output, the gene identifier
               (gene name or refseq id) are replaced by the gene symbol specified
                in the kgxref file. (If not found in the kgxref file, the
               gene identifiers are still used).</text>
          </comment>
        </parameter>
      </parameters>
    </paragraph>
    <paragraph>
      <name>query-db-match</name>
      <prompt lang="en">Criteria for defining query-db match</prompt>
      <parameters>
        <parameter>
          <name>condense_query</name>
          <prompt lang="en">Condense and eliminate overlapping regions in query</prompt>
          <type>
            <datatype>
              <class>Boolean</class>
            </datatype>
          </type>
          <vdef>
            <value>0</value>
          </vdef>
          <format>
            <code proglang="perl">($value) ? "--condense_query " : ""</code>
            <code proglang="python">( "" , " --condense_query " )[ value]</code>
          </format>
          <argpos>3</argpos>
          <comment>
            <text lang="en">Condense overlapped regions in the query file into non-over-lapped
         regions. When this argument is set, the annotation for
               each query (the strings after the chromosome location in each
               line of the query file) will not in the output.</text>
          </comment>
        </parameter>
        <parameter>
          <name>score_threshold</name>
          <prompt lang="en">Score threshold for database in UCSC annotation file</prompt>
          <type>
            <datatype>
              <class>Float</class>
            </datatype>
          </type>
          <format>
            <code proglang="perl">(defined $value) ? " --score_threshold $value " : ""</code>
            <code proglang="python">( "" , " --score_threshold " + str(value) )[ value is not None]</code>
          </format>
          <argpos>3</argpos>
          <comment>
            <text lang="en">Specify the score threshold in the database file to include in the
               search for overlaps. This argument is file format dependent.</text>
          </comment>
        </parameter>
        <parameter>
          <name>normscore_threshold</name>
          <prompt lang="en">Normalized score threshold for database in UCSC annotation file</prompt>
          <type>
            <datatype>
              <class>Float</class>
            </datatype>
          </type>
          <format>
            <code proglang="perl">(defined $value) ? " --normscore_threshold $value " : ""</code>
            <code proglang="python">( "" , " --normscore_threshold " + str(value) )[ value is not None]</code>
          </format>
          <argpos>3</argpos>
          <comment>
            <text lang="en">Specify the normalized score threshold in the database file to
               include in the search for overlaps. This argument is file format dependent.</text>
          </comment>
        </parameter>
        <!--
      <parameter>
        <name>minoverlap</name>
        <prompt lang="en">Minimum portion of overlap (either query or database) to decide query-db match</prompt>
        <type>
        <datatype>
          <class>Float</class>
        </datatype>
        </type>
        <format>
        <code proglang="perl">($value) ? " -minoverlap $value " : ""</code>
        <code proglang="python">( "" , " -minoverlap " + str(value) )[ value is not None]</code>
      </format>
      <argpos>3</argpos>
      <comment>
        <text lang="en">The minimum percentage of overlap between query and database chromosome region to infer a matching event. By default, even a sin-
               gle base pair overlap will be considered as overlap, but a lot
               of times people prefer to use something like 0.5 (50% overlap)
               to make sure that the query and database regions have high concordance. This argument operates by OR operation: if overlap for
               either query or database exceed the specified criteria, it will be
               treated as passing the criteria.</text>
      </comment>
      </parameter>
      
      <parameter>
        <name>maxoverlap</name>
        <prompt lang="en">Maximum portion of overlap (either query or database) to decide query-db match</prompt>
        <type>
        <datatype>
          <class>Float</class>
        </datatype>
        </type>
        <format>
        <code proglang="perl">($value) ? " -maxoverlap $value " : ""</code>
        <code proglang="python">( "" , " -maxoverlap " + str(value) )[ value is not None]</code>
      </format>
      <argpos>3</argpos>
      <comment>
        <text lang="en">The maximum percentage of overlap between query and database chromosome region to infer a matching event.</text>
      </comment>
      </parameter>           
                   
      <parameter>
        <name>minquerydbratio</name>
        <prompt lang="en">Minimum query to database length ratio to decide match</prompt>
        <type>
        <datatype>
          <class>Float</class>
        </datatype>
        </type>
        <format>
        <code proglang="perl">($value) ? " -minquerydbratio $value " : ""</code>
        <code proglang="python">( "" , " -minquerydbratio " + str(value) )[ value is not None]</code>
      </format>
      <argpos>3</argpos>
      <comment>
        <text lang="en">Minimum query to database length ratio to infer a matching event.</text>
      </comment>
      </parameter>
      
      <parameter>
        <name>maxquerydbratio</name>
        <prompt lang="en">Maximum query to database length ratio to decide match</prompt>
        <type>
        <datatype>
          <class>Float</class>
        </datatype>
        </type>
        <format>
        <code proglang="perl">($value) ? " -maxquerydbratio $value " : ""</code>
        <code proglang="python">( "" , " -maxquerydbratio " + str(value) )[ value is not None]</code>
      </format>
      <argpos>3</argpos>
      <comment>
        <text lang="en">Maximum query to database length ratio to infer a matching event.</text>
      </comment>
      </parameter>
      
      <parameter>
        <name>mindbfrac</name>
        <prompt lang="en">Minimum fraction of database in overlap</prompt>
        <type>
        <datatype>
          <class>Float</class>
        </datatype>
        </type>
        <format>
        <code proglang="perl">($value) ? " -mindbfrac $value " : ""</code>
        <code proglang="python">( "" , " -mindbfrac " + str(value) )[ value is not None]</code>
      </format>
      <argpos>3</argpos>
      <comment>
        <text lang="en">Minimum fraction of database in overlap to infer a matching event.</text>
      </comment>
      </parameter>
      
      <parameter>
        <name>maxdbfrac</name>
        <prompt lang="en">Maximum fraction of database in overlap</prompt>
        <type>
        <datatype>
          <class>Float</class>
        </datatype>
        </type>
        <format>
        <code proglang="perl">($value) ? " -maxdbfrac $value " : ""</code>
        <code proglang="python">( "" , " -maxdbfrac " + str(value) )[ value is not None]</code>
      </format>
      <argpos>3</argpos>
      <comment>
        <text lang="en">Maximum fraction of database in overlap to infer a matching event.</text>
      </comment>
      </parameter>
      
      <parameter>
        <name>minqueryfrac</name>
        <prompt lang="en">Minimum fraction of query in overlap</prompt>
        <type>
        <datatype>
          <class>Float</class>
        </datatype>
        </type>
        <format>
        <code proglang="perl">($value) ? " -minqueryfrac $value " : ""</code>
        <code proglang="python">( "" , " -minqueryfrac " + str(value) )[ value is not None]</code>
      </format>
      <argpos>3</argpos>
      <comment>
        <text lang="en">Minimum fraction of query in overlap to infer a matching event.</text>
      </comment>
      </parameter>
                   
      <parameter>
        <name>maxqueryfrac</name>
        <prompt lang="en">Maximum fraction of query in overlap</prompt>
        <type>
        <datatype>
          <class>Float</class>
        </datatype>
        </type>
        <format>
        <code proglang="perl">($value) ? " -maxqueryfrac $value " : ""</code>
        <code proglang="python">( "" , " -maxqueryfrac " + str(value) )[ value is not None]</code>
      </format>
      <argpos>3</argpos>
      <comment>
        <text lang="en">Maximum fraction of query in overlap to infer a matching event.</text>
      </comment>
      </parameter>
    
    -->
      </parameters>
    </paragraph>
    <paragraph>
      <name>expansion_query</name>
      <prompt lang="en">Expansion of query to find match</prompt>
      <parameters>
        <parameter>
          <name>expandleft</name>
          <prompt lang="en">Expand left side of query regions (overwrite --expandmax)</prompt>
          <type>
            <datatype>
              <class>Integer</class>
            </datatype>
          </type>
          <precond>
            <code proglang="perl">($reference eq '--knowngene' or $reference eq '--refgene') and (not $expandmax)</code>
            <code proglang="python">(reference == '--knowngene' or reference == '--refgene') and (not expandmax)</code>
          </precond>
          <format>
            <code proglang="perl">(defined $value) ? " --expandleft $value " : ""</code>
            <code proglang="python">( "" , " --expandleft " + str(value) )[ value is not None]</code>
          </format>
          <argpos>3</argpos>
          <comment>
            <text lang="en">Expand the query region on the left side (5 megabases in forward strand,
               3 megabases in reverse strand) to find overlap (used in conjunction with
               --refgene or --knowngene argument).</text>
          </comment>
        </parameter>
        <parameter>
          <name>expandright</name>
          <prompt lang="en">Expand right side of query regions (overwrite --expandmax)</prompt>
          <type>
            <datatype>
              <class>Integer</class>
            </datatype>
          </type>
          <precond>
            <code proglang="perl">($reference eq '--knowngene' or $reference eq '--refgene') and (not $expandmax)</code>
            <code proglang="python">(reference == '--knowngene' or reference == '--refgene') and (not expandmax)</code>
          </precond>
          <format>
            <code proglang="perl">(defined $value) ? " --expandright $value " : ""</code>
            <code proglang="python">( "" , " --expandright " + str(value) )[ value is not None]</code>
          </format>
          <argpos>3</argpos>
          <comment>
            <text lang="en">Expand the query region on the right side (3 megabases in forward
               strand, 5 megabases in reverse strand) to find overlap (used in conjunction
                with --refgene or --knowngene argument).</text>
          </comment>
        </parameter>
        <parameter>
          <name>expandmax</name>
          <prompt lang="en">Size of maximum expansion for query region to find overlap</prompt>
          <type>
            <datatype>
              <class>Integer</class>
            </datatype>
          </type>
          <precond>
            <code proglang="perl">$reference eq '--knowngene' or $reference eq '--refgene'</code>
            <code proglang="python">reference == '--knowngene' or reference == '--refgene'</code>
          </precond>
          <format>
            <code proglang="perl">(defined $value) ? " --expandmax $value " : ""</code>
            <code proglang="python">( "" , " --expandmax " + str(value) )[ value is not None]</code>
          </format>
          <argpos>3</argpos>
          <comment>
            <text lang="en">Maximum expansion size of the query region on both side to find
               at least one overlap (used in junction with --refgene or
               --knowngene argument). After query expansion, only the closet
               gene will be printed; other genes, even if overlapping with the
               query after expansion, will not be printed.</text>
          </comment>
        </parameter>
      </parameters>
    </paragraph>
    <parameter>
      <name>expanddb</name>
      <prompt lang="en">Expand definition of gene/cds/exon at both sides</prompt>
      <type>
        <datatype>
          <class>Integer</class>
        </datatype>
      </type>
      <format>
        <code proglang="perl">(defined $value) ? " --expanddb $value " : ""</code>
        <code proglang="python">( "" , " --expanddb " + str(value) )[ value is not None]</code>
      </format>
      <argpos>3</argpos>
      <comment>
        <text lang="en">Expand the chromosome region specified in the database-file to find
               overlap with the query regions.</text>
      </comment>
    </parameter>
    <paragraph>
      <name>output_option</name>
      <prompt lang="en">Input/output options</prompt>
      <parameters>
        <parameter>
          <name>overlap</name>
          <prompt lang="en">Print overlapped portion of region only</prompt>
          <type>
            <datatype>
              <class>Boolean</class>
            </datatype>
          </type>
          <vdef>
            <value>0</value>
          </vdef>
          <format>
            <code proglang="perl">($value) ? "--overlap " : ""</code>
            <code proglang="python">( "" , " --overlap " )[ value]</code>
          </format>
          <argpos>3</argpos>
          <comment>
            <text lang="en">Instead of printing the query region, only print the overlapped
               portion of the query region and template region.</text>
          </comment>
        </parameter>
        <parameter>
          <name>dbregion</name>
          <prompt lang="en">Print database region (default is to print query region)</prompt>
          <type>
            <datatype>
              <class>Boolean</class>
            </datatype>
          </type>
          <vdef>
            <value>0</value>
          </vdef>
          <format>
            <code proglang="perl">($value) ? "--dbregion " : ""</code>
            <code proglang="python">( "" , " --dbregion " )[ value]</code>
          </format>
          <argpos>3</argpos>
          <comment>
            <text lang="en">Print the region in database file, rather than query file, when an
               overlapped hit is found.</text>
          </comment>
        </parameter>
        <parameter>
          <name>append</name>
          <prompt lang="en">Append extra information from annotation file to output</prompt>
          <type>
            <datatype>
              <class>Boolean</class>
            </datatype>
          </type>
          <vdef>
            <value>0</value>
          </vdef>
          <format>
            <code proglang="perl">($value) ? "--append " : ""</code>
            <code proglang="python">( "" , " --append " )[ value]</code>
          </format>
          <argpos>3</argpos>
          <comment>
            <text lang="en">Append the score and normscore for the overlapped template
               region to the output for database files downloaded as UCSC tables.</text>
          </comment>
        </parameter>
        <parameter>
          <name>queryinfo</name>
          <prompt lang="en">Force to print query info when Print database region is used</prompt>
          <type>
            <datatype>
              <class>Boolean</class>
            </datatype>
          </type>
          <precond>
            <code proglang="perl">$dbregion</code>
            <code proglang="python">dbregion</code>
          </precond>
          <vdef>
            <value>0</value>
          </vdef>
          <format>
            <code proglang="perl">($value) ? "--queryinfo " : ""</code>
            <code proglang="python">( "" , " --queryinfo " )[ value]</code>
          </format>
          <argpos>3</argpos>
        </parameter>
      </parameters>
    </paragraph>
    <parameter isstdout="1">
      <name>output_file</name>
      <prompt lang="en">Output file</prompt>
      <type>
        <datatype>
          <class>Cnv</class>
          <superclass>AbstractText</superclass>
        </datatype>
      </type>
      <filenames>
        <code proglang="perl">"scan_region.out"</code>
        <code proglang="python">"scan_region.out"</code>
      </filenames>
    </parameter>
  </parameters>
</program>
mobyle-programs 5.1.2-2 / var / lib / mobyle / programs / scan_region.xml
