Mercurial > repos > urgi-team > teiso

changeset 1:15d6811e6bf5 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Uploaded
author urgi-team
date Tue, 24 May 2016 08:59:28 -0400
parents 3d22562b4489
children 775f119b9ac2
files TEiso/ClosestToStartSite_Wrapper.py TEiso/ClosestToStartSite_Wrapper.xml TEiso/CufflinksGTFToBed_Wrapper.py TEiso/CufflinksGTFToBed_Wrapper.xml TEiso/GFFToBed_Wrapper.py TEiso/GFFToBed_Wrapper.xml
diffstat 6 files changed, 125 insertions(+), 164 deletions(-) [+]
line wrap: on
line diff
--- a/TEiso/ClosestToStartSite_Wrapper.py	Fri Apr 29 09:11:18 2016 -0400
+++ b/TEiso/ClosestToStartSite_Wrapper.py	Tue May 24 08:59:28 2016 -0400
@@ -3,7 +3,6 @@
 
 import subprocess, tempfile, sys, os, glob, shutil, time
 from optparse import OptionParser
-from commons.core.utils.RepetOptionParser import RepetOptionParser
 
 
 class ClosestToStartSiteWrapper(object):
@@ -18,14 +17,16 @@
         
         
     def setAttributesFromCmdLine(self):
-        self._toolVersion = "1.0"
-        description = "ClosestToStartSite version %s" % self._toolVersion
-        epilog = "\n parses a bed file and create a bed file to create a report about positions of features A to features B. \n"
-        epilog += "example: ClosestToStartSite.py -i <inputFile> -o  <outputFile>\n"
-        parser = RepetOptionParser(description = description, epilog = epilog, version = self._toolVersion) 
-        parser.add_option("-i", "--inputFile",  dest = "inputFile",  action = "store", type = "string", help = "Input GTF File name(transcript.gtf of Cufflinks).",  default = "")
+        description = "ClosestToStartSite"
+        epilog = "\nParser a bed file and create a bed file to create a report about positions of features A to features B. \n"
+        epilog +="it can also add the class code of features A. \n"
+        epilog += "example: ClosestToStartSite.py -i <inputFile> -c <cuff_in.tmap> -o <outputFile>\n"
+        parser = OptionParser(description = description, version = "1.0")
+        parser.add_option("-i", "--inputFile",  dest = "inputFile",  action = "store", type = "string", help = "input bed file",  default = "")
+        parser.add_option("-c", "--cuffcom_tmap",  dest = "cuffcom_tmap",  action = "store", type = "string", help = "input gtf file",  default = "")
         parser.add_option("-o", "--outputFile", dest = "outputFile", action = "store", type = "string", help = "output Bed File name", default = "")
-        parser.add_option("-v", "--verbosity",  dest = "verbosity",  action = "store", type = "int",    help = "Verbosity [optional] [default: 3]",default = 3)
+        #parser.add_option("-t", "--outputFileclasscode", dest = "outputFile_classcode", action = "store", type = "string", help = "output Bed File name with class code.", default = "")
+        parser.add_option("-v", "--verbosity",  dest = "verbosity",  action = "store", type = "int",    help = "verbosity [optional] [default: 3]",default = 3)
         options = parser.parse_args()[0]
         self._setAttributesFromOptions(options)
 
@@ -36,6 +37,11 @@
         prg = "ClosestToStartSite.py"
         args = ""
         args += "-i %s" % self._options.inputFile
+	args += " "
+        args += "-o %s" % self._options.outputFile
+        if  self._options.cuffcom_tmap != "":
+            args += " "
+            args += "-c %s" % self._options.cuffcom_tmap
         cmd = "%s %s" %(prg, args)
         print cmd
         
--- a/TEiso/ClosestToStartSite_Wrapper.xml	Fri Apr 29 09:11:18 2016 -0400
+++ b/TEiso/ClosestToStartSite_Wrapper.xml	Tue May 24 08:59:28 2016 -0400
@@ -1,132 +1,69 @@
 <tool id="ClosestToStartSite" name="ClosestToStartSite" version="1.0">
-    <description>ClosestToStartSite parses a bed file and create a bed file to create a report about positions of features A to features B.</description>
+    <description>ClosestToStartSite parses a bed file and generate a report about positions of features A to features B. It can also add the class code of features A.</description>
     <requirements>
         <requirement type="package" version="1.0">TEiso_Tools</requirement>
     </requirements>
     <version_command>
    ClosestToStartSite.py --version
     </version_command>
-    <command interpreter="python">
-    ClosestToStartSite_Wrapper.py -i $inputFile -o $outputFile
+   <command interpreter="python">
+    #if $ClassCode.get_class_code
+    	ClosestToStartSite_Wrapper.py -i $inputFile -c $cuffcom_tmap -o $outputFile
+    #else
+    	ClosestToStartSite_Wrapper.py -i $inputFile -o $outputFile
+     #end if
     </command>
     <inputs>
         <param name="inputFile" type="data" format="bed" label="indicate a bed file."/>
+        <conditional name="ClassCode">
+            <param name="get_class_code" type="boolean" label="get information of class code from cuff_in.tmap" truevalue="yes" falsevalue="no" checked="off" />
+            <when value="no"/>
+            <when value="yes">
+                <param name="cuffcom_tmap" type="data" format="tabular" label="indicate cuff_in.tmap" help="default = 1"/>
+            </when>
+        </conditional>
     </inputs>
     <outputs>
         <data  name="outputFile" format="bed" label="${tool.name} on ${on_string} (BED)"/>
     </outputs>
     <help><![CDATA[
       
-**ClosestToStartSite_wrapper parses a bed file and create a bed file to create a report about positions of features A to features B.**
-
-
-**what it does :**
-
-parses a bed file and create a bed file to create a report about positions of features A to features B
-
------
-
-**input format :**
-
-.. class:: infomark
-
-**output format :**
-
-fake	140	532	CUFF.1.1	CUFF.1	+	26875.607
-
------
 
 
 
-******  "B_close_TSS"
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+**ClosestToStartSite parses a bed file and generate a report about positions of features A (according to its TSS) to features B. It can also add the class code of features A.**
+
+**what it does :**
 
-					              F[1]   gene             F[2]
-                                    =========================>
-                     ------------
-                    F[8]        F[9]
-    
-    
-                                   F[1]                     F[2]
-                                    <=========================
-                                                                ---------------
+generate a report about positions of features B to features A (according to its TSS). In input file, there are the information of the two features. ClosestToStartSite reports only the cases that features Bis near to TSS or has overlap with TSS of features A. it can be also run cuffcopare to features A, and add the class code information to output file.
+
+
+-----
+
+**example :**
 
 
-******  "B_overlap_TSS"
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
-            
-                                    F[1]   gene             F[2]
-                                    =========================>
-                             -------------
-                            F[8]        F[9]
-            
-                                                     gene
-                 					F[1]=========================>F[2]
-            
-      								F[8]---------------F[9]
-            
-               
-            
-                                   F[1]=============================>F[2]
-                F[8]---------------F[9]
-            
-            
-                                    F[1]<=============================F[2]
-                                                       ---------------------------
-                                                    F[8]           F[9]
-            
-            
-                                F[1]<=============================F[2]
-                                               F[8]---------------F[9]
-            
-                  
-                                                            F[1]<=============================F[2]
-                                                                                            F[8]---------------F[9]
+Bed input file: ::
+
+    2L  239751  240346  CUFF.36.1       CUFF.36 -       2L      239932  241306  RXX     +       415     TE_overlap_TSS
+    2L  113233  113411  CUFF.14.1       CUFF.14 -       2L      113496  113509  RLX     -       -854    TE_near_TSS
+
+
+----
+
+output format: ::
+
 
-******  "B-inclus-A"
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
-            
-                    
-            
-                                    F[1]            gene          F[2]
-                                    ==============================
-                                            -------------
-                                            F[8]        F[9]
-            
-******  "A-inclus-B"
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~            
-            
-                        
-                                    F[1]======================F[2]
-                    F[8]----------------------------------------------------F[9]
-            
-            
-            
-            
-         					         F[1]==================================>F[2]
-             F[8]----------------------------------------------------------F[9]
-             
-             
-             
-                 					F[1]<==================================F[2]
-                    F[8]----------------------------------------------------------F[9]
-             
-             
-            
-******  "A-inclus-B-inTSS"
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~           
-                    
- 											 F[1]<==================================F[2]
-                       [8]----------------------------------------------------------F[9]
-            
-                  
- 												F[1]==================================>F[2]
-                                   					F[8]----------------------------------------------------------F[9]
-            
-    
------
+    2L	239751	240346	CUFF.36.1	CUFF.36	-	2L	239932	241306	RXX	+	415	TE_overlap_TSS
+    2L	113233	113411	CUFF.14.1	CUFF.14	-	2L	113496	113509	RLX	-	-854	TE_near_TSS
+
 
-**reference :**
+output format with calss code: ::
+ 
+
+    2L	239751	240346	CUFF.36.1	CUFF.36	-	2L	239932	241306	RXX	+	415	TE_overlap_TSS	c	gene-id-1
+    2L	113233	113411	CUFF.14.1	CUFF.14	-	2L	113496	113509	RLX	-	-854	TE_near_TSS	=	gene-id-2
+
 
 ]]>
     </help>
--- a/TEiso/CufflinksGTFToBed_Wrapper.py	Fri Apr 29 09:11:18 2016 -0400
+++ b/TEiso/CufflinksGTFToBed_Wrapper.py	Tue May 24 08:59:28 2016 -0400
@@ -1,9 +1,7 @@
 #!/usr/bin/env python
-
-
-import subprocess, tempfile, sys, os, glob, shutil, time
+import subprocess, tempfile, sys, os
 from optparse import OptionParser
-from commons.core.utils.RepetOptionParser import RepetOptionParser
+#from commons.core.utils.RepetOptionParser import RepetOptionParser
 
 
 class CufflinksGTFToBedWrapper(object):
@@ -18,11 +16,12 @@
         
         
     def setAttributesFromCmdLine(self):
-        self._toolVersion = "1.0"
-        description = "CufflinksGTFToBed version %s" % self._toolVersion
+        #self._toolVersion = "1.0"
+        description = "CufflinksGTFToBed "
         epilog = "\n parses a GTF file of Cufflinks and create a bed file. \n"
         epilog += "example: CufflinksGTFToBed.py -i <inputFile> -o  <outputFile>\n"
-        parser = RepetOptionParser(description = description, epilog = epilog, version = self._toolVersion) 
+        #parser = RepetOptionParser(description = description, epilog = epilog, version = self._toolVersion) 
+        parser = OptionParser(description = description, version = "1.0")
         parser.add_option("-i", "--inputFile",  dest = "inputFile",  action = "store", type = "string", help = "Input GTF File name(transcript.gtf of Cufflinks).",  default = "")
         parser.add_option("-o", "--outputFile", dest = "outputFile", action = "store", type = "string", help = "output Bed File name", default = "")
         parser.add_option("-v", "--verbosity",  dest = "verbosity",  action = "store", type = "int",    help = "Verbosity [optional] [default: 3]",default = 3)
@@ -33,9 +32,12 @@
         self._options = options
 
     def run(self):
+	tmp = "%s_tmp" % ((os.path.splitext(self._options.outputFile)[0]))
         prg = "CufflinksGTFToBed.py"
         args = ""
         args += "-i %s" % self._options.inputFile
+	args += " "
+        args += "-o %s" % tmp
         cmd = "%s %s" %(prg, args)
         print cmd
         
@@ -61,6 +63,12 @@
                 raise Exception, stderr
         except Exception, e:
             self.stop_err( 'Error in TranscriptToBed:\n' + str( e ) ) 
+	try:
+            cmdsort= "bedtools sort -i %s > %s" % (tmp, self._options.outputFile)
+            os.system(cmdsort)
+        except Exception, e:
+            self.stop_err( 'Error in bedtools sort:\n' + str( e ) )
+
         
 if __name__ == "__main__":
     iWrapper = CufflinksGTFToBedWrapper()
--- a/TEiso/CufflinksGTFToBed_Wrapper.xml	Fri Apr 29 09:11:18 2016 -0400
+++ b/TEiso/CufflinksGTFToBed_Wrapper.xml	Tue May 24 08:59:28 2016 -0400
@@ -16,36 +16,35 @@
         <data  name="outputFile" format="bed" label="${tool.name} on ${on_string} (BED)"/>
     </outputs>
     <help><![CDATA[
-      
+
+
+
 **CufflinksGTFToBed_wrapper converts a result GTF file of Cufflinks into a bed file.**
 
-
 **what it does :**
 
-converts a result GTF file of Cufflinks into a bed file. 
-
+Converts a GTF file of transcripts into a bed file. In input file, there are all the information of the transcript and their exon. CufflinksGTFToBed takes only the transcripts to convert into a bed file. 
+ 
 It can take: Chromosome, Start, End, strand, Isoform ID, Gene ID, value of FPKM
 
 -----
 
-**input format :**
+**example :**
 
-.. class:: infomark
 
-fake	Cufflinks	transcript	140	532	1000	+	.	gene_id "CUFF.1"; 
-transcript_id "CUFF.1.1"; FPKM "26875.6073354154"; frac "1.000000"; conf_lo "24989.806681"; conf_hi "28761.407990"; cov "752.375132";
------
+Transcript input file: ::
 
-**output format :**
-
-fake	140	532	CUFF.1.1	CUFF.1	+	26875.607
-
------
+    2L	cufflinks	transcript	10487	18076	372	-	.	gene_id "CUFF.1"; transcript_id "CUFF.1.1"; FPKM "1.3627628649"; frac "0.197041"; conf_lo "0.805777"; conf_hi "1.919748"; cov "1.706083";
 
 
------
+      
+
+----
 
-**reference :**
+output format: ::
+
+    2L	10487	18076	CUFF.1.1	CUFF.1	-	1.363
+
 
 ]]>
     </help>
--- a/TEiso/GFFToBed_Wrapper.py	Fri Apr 29 09:11:18 2016 -0400
+++ b/TEiso/GFFToBed_Wrapper.py	Tue May 24 08:59:28 2016 -0400
@@ -3,7 +3,7 @@
 
 import subprocess, tempfile, sys, os, glob, shutil, time
 from optparse import OptionParser
-from commons.core.utils.RepetOptionParser import RepetOptionParser
+#from commons.core.utils.RepetOptionParser import RepetOptionParser
 
 
 class GFFToBedWrapper(object):
@@ -18,12 +18,13 @@
         
         
     def setAttributesFromCmdLine(self):
-        self._toolVersion = "1.0"
-        description = "GFFToBed version %s" % self._toolVersion
-        epilog = "\n parses a GFF file and create a bed file. \n"
+       ## self._toolVersion = "1.0"
+        description = "GFFToBed version"
+        epilog = "\n parses a GFF3 file and create a bed file. \n"
         epilog += "example: GFFToBed.py -i <inputFile> -o  <outputFile>\n"
-        parser = RepetOptionParser(description = description, epilog = epilog, version = self._toolVersion) 
-        parser.add_option("-i", "--inputFile",  dest = "inputFile",  action = "store", type = "string", help = "Input GFF File name.",  default = "")
+        #parser = RepetOptionParser(description = description, epilog = epilog, version = self._toolVersion)
+        parser = OptionParser(description = description, version = "1.0") 
+        parser.add_option("-i", "--inputFile",  dest = "inputFile",  action = "store", type = "string", help = "Input GFF3 File name.",  default = "")
         parser.add_option("-o", "--outputFile", dest = "outputFile", action = "store", type = "string", help = "output Bed File name", default = "")
         parser.add_option("-v", "--verbosity",  dest = "verbosity",  action = "store", type = "int",    help = "Verbosity [optional] [default: 3]",default = 3)
         options = parser.parse_args()[0]
@@ -33,12 +34,15 @@
         self._options = options
 
     def run(self):
-        prg = "GFFToBed.py"
+	tmp = "%s_tmp" % ((os.path.splitext(self._options.outputFile)[0]))
+	prg = "GFFToBed.py"
         args = ""
         args += "-i %s" % self._options.inputFile
+        args += " "
+        args += "-o %s" % tmp
         cmd = "%s %s" %(prg, args)
         print cmd
-        
+ 
         try:
             tmp_err = tempfile.NamedTemporaryFile().name
             tmp_stderr = open( tmp_err, 'wb' )
@@ -61,7 +65,16 @@
                 raise Exception, stderr
         except Exception, e:
             self.stop_err( 'Error in GFFToBed:\n' + str( e ) ) 
-        
+
+	
+	try:
+	    cmdsort= "bedtools sort -i %s > %s" % (tmp, self._options.outputFile)
+	    os.system(cmdsort)
+        except Exception, e:
+            self.stop_err( 'Error in bedtools sort:\n' + str( e ) )
+
+
+
 if __name__ == "__main__":
     iWrapper = GFFToBedWrapper()
     iWrapper.setAttributesFromCmdLine()
--- a/TEiso/GFFToBed_Wrapper.xml	Fri Apr 29 09:11:18 2016 -0400
+++ b/TEiso/GFFToBed_Wrapper.xml	Tue May 24 08:59:28 2016 -0400
@@ -1,5 +1,5 @@
 <tool id="GFFToBed" name="GFFToBed" version="1.0">
-    <description>GFFToBed can convert a result GTF file of Cufflinks into a bed file.</description>
+    <description>GFFToBed can convert a result GTF file into a bed file.</description>
     <requirements>
         <requirement type="package" version="1.0">TEiso_Tools</requirement>
     </requirements>
@@ -7,7 +7,7 @@
    GFFToBed.py --version
     </version_command>
     <command interpreter="python">
-    GFFToBedWrapper_Wrapper.py -i $inputFile -o $outputFile
+    GFFToBed_Wrapper.py -i $inputFile -o $outputFile
     </command>
     <inputs>
         <param name="inputFile" type="data" format="gff" label="indicate a transcript GTF file of cufflinks."/>
@@ -16,37 +16,35 @@
         <data  name="outputFile" format="bed" label="${tool.name} on ${on_string} (BED)"/>
     </outputs>
     <help><![CDATA[
-      
-**GFFToBed_wrapper converts a result GTF file of Cufflinks into a bed file.**
+
 
+**GFFToBed_wrapper converts a result GTF file into a bed file.**
 
 **what it does :**
 
-converts a result GTF file of Cufflinks into a bed file. 
-
-It can take: Chromosome, Start, End, ID, Target, strand
+converts a result GFF3 file into a bed file
+ 
+it can take: Chromosome, Start, End, ID, Target, strand.
 
 -----
 
-**input format :**
+**example :**
+
+a GFF3 file of the Transposable elements as input file: ::
 
-.. class:: infomark
-
-2L	DmelCaf1_2_2_REPET_TEs	match	47519	52563	0.0	+	.	ID=ms1_dmel_chr2L_RIX-comp_DmelCaf1_2_2-B-P58.20-Map11;Target=RIX-comp_DmelCaf1_2_2-B-P58.20-Map11 6 5050;Identity=99.8
-
+    2L	DmelCaf1_2_2_REPET_TEs	match	47519	52563	0.0	+	.	ID=ms1_dmel_chr2L_RIX-comp_DmelCaf1_2_2-B-P58.20-Map11;Target=RIX-comp_DmelCaf1_2_2-B-P58.20-Map11 6 5050;Identity=99.8   
+    2L	DmelCaf1_2_2_REPET_TEs	match	16205	16490	0.0	+	.	ID=ms2_dmel_chr2L_RXX-LARD_DmelCaf1_2_2-L-B578-Map1;Target=RXX-LARD_DmelCaf1_2_2-L-B578-Map1 2551 2828;Identity=97.5
+    2R	DmelCaf1_2_2_REPET_TEs	match	24134   24428   0.0	-	.	ID=ms4350_dmel_chr2R_RXX-LARD-chim_DmelCaf1_2_2-L-B414-Map1;Target=RXX-LARD-chim_DmelCaf1_2_2-L-B414-Map1 2894 3273
 
 -----
 
-**output format :**
+output format: ::
 
-2L	47519	52563	ms1_dmel_chr2L_RIX-comp_DmelCaf1_2_2-B-P58.20-Map11		RIX-comp_DmelCaf1_2_2-B-P58.20-Map11 6 5050		+
-
------
+    2L	47519	52563	ms1_dmel_chr2L_RIX-comp_DmelCaf1_2_2-B-P58.20-Map11		RIX-comp_DmelCaf1_2_2-B-P58.20-Map11 6 5050	+
+    2L	16205	16490	ms2_dmel_chr2L_RXX-LARD_DmelCaf1_2_2-L-B578-Map1		RXX-LARD_DmelCaf1_2_2-L-B578-Map1 2551 2828	+
+    2R	24134	24428	ms4350_dmel_chr2R_RXX-LARD-chim_DmelCaf1_2_2-L-B414-Map1	RXX-LARD-chim_DmelCaf1_2_2-L-B414-Map1 2894	3273	-
 
 
------
-
-**reference :**
 
 ]]>
     </help>