Changeset 5829

Timestamp:

Feb 19, 2020, 3:21:30 PM (3 years ago)

Author:

Nicklas Nordborg

Message:

References #1218: Implement MIPs alignment

Added AnnotateBamWithUmis step.

Location:

extensions/net.sf.basedb.reggie/trunk

Files:

• config/reggie-config.xml (modified) (2 diffs)
• src/net/sf/basedb/reggie/grid/MipsAlignJobCreator.java (modified) (4 diffs)

extensions/net.sf.basedb.reggie/trunk/config/reggie-config.xml

@@ -198 +198 @@
           <path>/usr/local/packages/snpeff/4.3s/SnpSift.jar</path>
         </snpsift>
+        <fgbio>
+          <!-- full path to the fgbio.jar file (required) -->
+          <path>/usr/local/packages/fgbio/0.8.1/fgbio.jar</path>
+        </fgbio>
       </programs>
       
@@ -335 +339 @@
         </trimmomatic>
         
+        
       </align-mips>
       

extensions/net.sf.basedb.reggie/trunk/src/net/sf/basedb/reggie/grid/MipsAlignJobCreator.java

@@ -143 +143 @@
     }
     String picard_path = cfg.getRequiredConfig("programs/picard/path", alignParameterSet);
+    String fgbio_path = cfg.getRequiredConfig("programs/fgbio/path", alignParameterSet);
 
     // Options for the programs
@@ -258 +259 @@
       script.cmd("JAVA="+java_path);
       script.cmd("TrimmomaticJAR="+trimmomatic_path);
+      script.cmd("FgBioJAR="+fgbio_path);
       script.cmd("ArchiveRoot="+archiveRoot);
       script.cmd("FastqFolder=${ArchiveRoot}"+fastQFolder);
@@ -281 +283 @@
       script.cmd("mkdir trimmomatic.2");
       script.cmd("mkdir bam.1");
+      script.cmd("mkdir bam.2");
       script.cmd("cp ${ScriptDir}/stdwrap.sh .");
       script.cmd("cp ${ScriptDir}/mips_adapters.sh .");
@@ -337 +340 @@
       script.cmd("done");
       
-      script.comment("Run Picard FastqToSam");
+      script.comment("Run Picard FastqToSam and annotat the BAM files with UMI");
       script.cmd("for prefix in ${FastqPrefix[@]} ; do");
       // Picard2 uses different parameter syntax, but the .rg file is stored in picard1 syntax
       script.cmd("   rg_info=`./convert_picard_parameters.sh < fastq/${prefix}.rg`");
 
-      String toSamCmd = "./stdwrap.sh ./picard2 FastqToSam";
-      toSamCmd += " -SORT_ORDER queryname";
-      toSamCmd += " -QUALITY_FORMAT Standard";
-      toSamCmd += " ${rg_info}";
-      toSamCmd += " -FASTQ trimmomatic.2/${prefix}_R1.fastq";
-      toSamCmd += " -FASTQ2 trimmomatic.2/${prefix}_R2.fastq";
-      toSamCmd += " -OUTPUT bam.1/${prefix}.bam";
-      toSamCmd += " >> fastqtosam.out";
-      script.cmd("   " + toSamCmd);
+      String toBamCmd = "./stdwrap.sh ./picard2 FastqToSam";
+      toBamCmd += " -SORT_ORDER queryname";
+      toBamCmd += " -QUALITY_FORMAT Standard";
+      toBamCmd += " ${rg_info}";
+      toBamCmd += " -FASTQ trimmomatic.2/${prefix}_R1.fastq";
+      toBamCmd += " -FASTQ2 trimmomatic.2/${prefix}_R2.fastq";
+      toBamCmd += " -OUTPUT bam.1/${prefix}.bam";
+      toBamCmd += " >> fastqtosam.out";
+      
+      String umiCmd = "./stdwrap.sh ${JAVA} -jar ${FgBioJAR} AnnotateBamWithUmis";
+      umiCmd += " --fail-fast true -t RX";
+      umiCmd += " -i bam.1/${prefix}.bam";
+      umiCmd += " -f fastq/${prefix}_UMI.fastq.gz";
+      umiCmd += " -o bam.2/${prefix}.bam";
+      umiCmd += " >> annotatebam.out";
+      
+      script.cmd("   " + toBamCmd);
+      script.cmd("   " + umiCmd);
       script.cmd("done");
       /*