Changeset 5828

Timestamp:

Feb 19, 2020, 2:02:00 PM (3 years ago)

Author:

Nicklas Nordborg

Message:

References #1218: Implement MIPs alignment

Added FastqToSam? step that convertes paired FASTQ files to BAM format.

File:

• extensions/net.sf.basedb.reggie/trunk/src/net/sf/basedb/reggie/grid/MipsAlignJobCreator.java (modified) (6 diffs)

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

@@ -142 +142 @@
       trimmomatic_path = cfg.getRequiredConfig("programs/trimmomatic/path", alignParameterSet);
     }
+    String picard_path = cfg.getRequiredConfig("programs/picard/path", alignParameterSet);
 
     // Options for the programs
@@ -254 +255 @@
       script.comment("Setting up scripting environment and copying script to tmp folder");
       script.cmd("ScriptDir=" + pipeline_scripts_path);
+      script.cmd("PicardDir="+picard_path);
       script.cmd("JAVA="+java_path);
       script.cmd("TrimmomaticJAR="+trimmomatic_path);
@@ -278 +280 @@
       script.cmd("mkdir trimmomatic.1");
       script.cmd("mkdir trimmomatic.2");
+      script.cmd("mkdir bam.1");
       script.cmd("cp ${ScriptDir}/stdwrap.sh .");
       script.cmd("cp ${ScriptDir}/mips_adapters.sh .");
+      script.cmd("cp ${ScriptDir}/picard2 .");
+      script.cmd("cp ${ScriptDir}/convert_picard_parameters.sh .");
       script.newLine();
 
@@ -286 +291 @@
       script.newLine();
       
-      script.comment("Copy FASTQ files to tmp folder");
+      script.comment("Copy FASTQ and .rg files to tmp folder");
       script.progress(10, "Copying FASTQ files");
       script.cmd("if [ ! -d \"${FastqFolder}\" ]; then");
@@ -293 +298 @@
       script.cmd("fi");
       script.cmd("cp ${FastqFolder}/*.fastq.gz fastq");
+      script.cmd("cp ${FastqFolder}/*.rg fastq");
       script.newLine();
 
@@ -328 +334 @@
       trimCmd2 += " >> trimmomatic.2/trimmomatic.out";
       script.cmd("   " + trimCmd1);
-      script.cmd("   " + trimCmd2);   
+      script.cmd("   " + trimCmd2);
       script.cmd("done");
       
+      script.comment("Run Picard FastqToSam");
+      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);
+      script.cmd("done");
       /*
       script.progress(95, "Copying result files to project archive");