Add CCAT ChIP-seq peak/region caller.

2026-09-24 16:30:27 +08:00 · 2011-01-03 11:15:29 -05:00
parent b2b891d52d
commit b9d790af55
5 changed files with 173 additions and 0 deletions
@@ -0,0 +1,4 @@
+#Provide Pre-compiled CCAT peak/region caller configuration files
+#UID	Description	path_to_config_file
+#ccat_2.0_tf_config	CCAT provided Transcription Factor Binding	/path_to/ccat/configs/config_TF.txt
+#ccat_2.0_histone_config	CCAT provided Histone Modification	/path_to/ccat/configs/config_histone.txt
@@ -313,6 +313,7 @@
  </section>
  <section name="NGS: Peak Calling" id="peak_calling">
   <tool file="peak_calling/macs_wrapper.xml" />
+   <tool file="peak_calling/ccat_2_wrapper.xml" />
   <tool file="genetrack/genetrack_indexer.xml" />
   <tool file="genetrack/genetrack_peak_prediction.xml" />
  </section>
@@ -70,4 +70,9 @@
        <columns>value, dbkey, name, path</columns>
        <file path="tool-data/bowtie_indices.loc" />
    </table>
+    <!-- Locations of configurations in the CCAT peak/region caller format -->
+    <table name="ccat_configurations" comment_char="#">
+        <columns>value, name, path</columns>
+        <file path="tool-data/ccat_configurations.loc" />
+    </table>
 </tables>
@@ -0,0 +1,40 @@
+import sys, subprocess, tempfile, shutil, os.path
+
+CCAT_BINARY = "CCAT"
+
+def get_top_count( filename ):
+    for line in open( filename ):
+        if line.startswith( 'outputNum' ):
+            return int( line.split()[-1].strip() ) 
+
+def stop_err( tmp_dir, exception ):
+    print >> sys.stderr, "Error running CCAT."
+    shutil.rmtree( tmp_dir ) #some error has occurred, provide info and remove possibly non-empty temp directory
+    raise exception
+
+def main():
+    input_tag_file = sys.argv[1]
+    input_control_file = sys.argv[2]
+    chrom_info_file = sys.argv[3]
+    input_config_file = sys.argv[4]
+    project_name = sys.argv[5]
+    output_peak_file = sys.argv[6]
+    output_region_file = sys.argv[7]
+    output_top_file = sys.argv[8]
+    output_log_file = sys.argv[9]
+    
+    tmp_dir = tempfile.mkdtemp()
+    try:
+        proc = subprocess.Popen( args="%s %s > %s" % ( CCAT_BINARY, " ".join( map( lambda x: "'%s'" % x, [ input_tag_file, input_control_file, chrom_info_file, input_config_file, project_name ] ) ), output_log_file ), shell=True, cwd=tmp_dir )
+        proc.wait()
+        if proc.returncode:
+            raise Exception( "Error code: %i" % proc.returncode )
+        output_num = get_top_count( input_config_file )
+        shutil.move( os.path.join( tmp_dir, "%s.significant.peak" % project_name ), output_peak_file )
+        shutil.move( os.path.join( tmp_dir, "%s.significant.region" % project_name ), output_region_file )
+        shutil.move( os.path.join( tmp_dir, "%s.top%i.peak" % ( project_name, output_num ) ), output_top_file )
+    except Exception, e:
+        return stop_err( tmp_dir, e )
+    os.rmdir( tmp_dir ) #clean up empty temp working directory
+
+if __name__ == "__main__": main()
@@ -0,0 +1,123 @@
+<tool id="peakcalling_ccat2" name="CCAT" version="0.0.1">
+  <description>Control-based ChIP-seq Analysis Tool</description>
+  <command interpreter="python">ccat_2_wrapper.py '$input_tag_file' '$input_control_file' '$chromInfo' 
+  #if str( $options_type[ 'options_type_selector' ] ) == 'advanced':
+  '$input_advanced_config_file' 
+  #else:
+  '${ options_type.input_config_file.get_field( 'path' ) }'
+  #end if
+  'CCAT in Galaxy' 
+  '$output_peak_file' '$output_region_file' '$output_top_file' '$output_log_file'</command>
+  <requirements>
+    <requirement type="binary">CCAT</requirement>
+  </requirements>
+  <inputs>
+    <param name="input_tag_file" type="data" format="bed" label="ChIP-Seq Tag File" >
+      <validator type="unspecified_build" />
+    </param>
+    <param name="input_control_file" type="data" format="bed" label="ChIP-Seq Control File" >
+      <validator type="unspecified_build" />
+    </param>
+    <conditional name="options_type">
+      <param name="options_type_selector" type="select" label="Advanced Options">
+        <option value="basic" selected="True">Hide Advanced Options</option>
+        <option value="advanced">Show Advanced Options</option>
+      </param>
+      <when value="basic">
+        <param name="input_config_file" type="select" label="Select a pre-defined configuration file">
+          <options from_data_table="ccat_configurations">
+            <validator type="no_options" message="No configurations are available"/>
+          </options>
+        </param>
+      </when>
+      <when value="advanced">
+        <param name="fragment_size" type="integer" label="Length of DNA fragment" value="200"/>
+        <param name="sliding_window_size" type="integer" label="Sliding window size" value="500" help="transcription factor binding default: 300; histone modifications default: 500"/>
+        <param name="moving_step" type="integer" label="Step of sliding window" value="50" help="transcription factor binding default: 10; histone modifications default: 50"/>
+        <param name="is_strand_sensitive_mode" type="select" label="isStrandSensitiveMode" >
+          <option value="1">Transition from sense strand to anti-sense strand</option>
+          <option value="0" selected="True">Local maximum of read-enrichment profile</option>
+        </param>
+        <param name="min_count" type="integer" label="Minimum number of read counts at the peak" value="4"/>
+        <param name="output_num" type="integer" label="Number of peaks reported in top peak file" value="100000"/>
+        <param name="random_seed" type="integer" label="Random Seed" value="123456"/>
+        <param name="min_score" type="float" label="Minimum score of normalized difference" value="3.0"/>
+        <param name="bootstrap_pass" type="integer" label="Number of passes in the bootstrapping process" value="50"/>
+      </when>
+    </conditional>
+  </inputs>
+  <outputs>
+    <data name="output_peak_file" format="interval" label="${tool.name} on ${on_string} (peaks)">
+      <actions>
+        <action type="metadata" name="chromCol" default="1"/>
+        <action type="metadata" name="startCol" default="3"/>
+        <action type="metadata" name="endCol" default="4"/>
+      </actions>
+    </data>
+    <data name="output_region_file" format="interval" label="${tool.name} on ${on_string} (regions)">
+      <actions>
+        <action type="metadata" name="chromCol" default="1"/>
+        <action type="metadata" name="startCol" default="3"/>
+        <action type="metadata" name="endCol" default="4"/>
+      </actions>
+    </data>
+    <data name="output_top_file" format="interval" label="${tool.name} on ${on_string} (top peaks)">
+      <actions>
+        <action type="metadata" name="chromCol" default="1"/>
+        <action type="metadata" name="startCol" default="3"/>
+        <action type="metadata" name="endCol" default="4"/>
+      </actions>
+    </data>
+    <data name="output_log_file" format="txt" label="${tool.name} on ${on_string} (log)"/>
+  </outputs>
+  <configfiles>
+    <configfile name="input_advanced_config_file">#if str( $options_type['options_type_selector' ] ) == 'advanced':
+fragmentSize	${options_type[ 'fragment_size' ]}
+slidingWinSize	${options_type[ 'sliding_window_size' ]}
+movingStep	${options_type[ 'moving_step' ]}
+isStrandSensitiveMode	${options_type[ 'is_strand_sensitive_mode' ]}
+minCount	${options_type[ 'min_count' ]}
+outputNum	${options_type[ 'output_num' ]}
+randomSeed	${options_type[ 'random_seed' ]}
+minScore	${options_type[ 'min_score' ]}	
+bootstrapPass	${options_type[ 'bootstrap_pass' ]}
+#end if</configfile>
+  </configfiles>
+  <tests>
+    <test>
+      <param name="input_tag_file" value="chipseq_enriched.bed.gz" ftype="bed" dbkey="hg18" />
+      <param name="input_control_file" value="chipseq_input.bed.gz" ftype="bed" dbkey="hg18" />
+      <param name="options_type_selector" value="basic" />
+      <param name="input_config_file" value="ccat_2.0_histone_config" />
+      <output name="output_peak_file" file="peakcalling_ccat2/ccat2_test_peak_out_1.interval" />
+      <output name="output_region_file" file="peakcalling_ccat2/ccat2_test_region_out_1.interval" />
+      <output name="output_top_file" file="peakcalling_ccat2/ccat2_test_top_out_1.interval" />
+      <output name="output_log_file" file="peakcalling_ccat2/ccat2_test_log_out_1.interval" />
+    </test>
+    <test>
+      <param name="input_tag_file" value="chipseq_enriched.bed.gz" ftype="bed" dbkey="hg18" />
+      <param name="input_control_file" value="chipseq_input.bed.gz" ftype="bed" dbkey="hg18" />
+      <param name="options_type_selector" value="advanced" />
+      <param name="fragment_size" value="200" />
+      <param name="sliding_window_size" value="500" />
+      <param name="moving_step" value="50" />
+      <param name="is_strand_sensitive_mode" value="0" />
+      <param name="min_count" value="4" />
+      <param name="output_num" value="100000" />
+      <param name="random_seed" value="123456" />
+      <param name="min_score" value="3.0" />
+      <param name="bootstrap_pass" value="50" />
+      <output name="output_peak_file" file="peakcalling_ccat2/ccat2_test_peak_out_1.interval" />
+      <output name="output_region_file" file="peakcalling_ccat2/ccat2_test_region_out_1.interval" />
+      <output name="output_top_file" file="peakcalling_ccat2/ccat2_test_top_out_1.interval" />
+      <output name="output_log_file" file="peakcalling_ccat2/ccat2_test_log_out_1.interval" />
+    </test>
+  </tests>
+  <help>
+**What it does**
+
+This tool allows ChIP-seq peak/region calling using CCAT.
+
+View the original CCAT documentation: http://cmb.gis.a-star.edu.sg/ChIPSeq/paperCCAT.htm.
+  </help>
+</tool>