Changeset 5835

Timestamp:

Feb 24, 2020, 1:37:16 PM (3 years ago)

Author:

Nicklas Nordborg

Message:

References #1218: Implement MIPs alignment

Re-designed the two steps that convert FASTQ files to BAM and annotate them with UMI information.

File:

• extensions/net.sf.basedb.reggie/trunk/src/net/sf/basedb/reggie/grid/MipsAlignJobCreator.java (modified) (1 diff)

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

@@ -340 +340 @@
       script.cmd("   run_trimmomatic \"${prefix}\" $n");
       script.cmd("done");
+      script.newLine();
       script.cmd("trimlog_stats \"tmp/*.adaptrim.log\" \"out/adaptrim.hist.txt\"");
       script.cmd("trimlog_stats \"tmp/*.qualtrim.log\" \"out/qualtrim.hist.txt\"");
+      script.newLine();
       script.cmd("if [ ${nAfterTrim} -eq 0 ]; then");
       script.cmd("  echo \"No paired end reads surviving adapter and quality trimming for any fastq prefix\" 1>&2");
       script.cmd("exit 1");
       script.cmd("fi");
-      
+      script.newLine();
+      
+      script.comment("Run Picard FastqToSam and annotate the BAM files with UMI");
+      script.progress(30, "Converting FASTQ to BAM and annotating with UMI");
+      script.cmd("n=0");
+      script.cmd("for prefix in ${FastqPrefix[@]} ; do");
+      script.cmd("   n=$((n + 1))");
+      script.cmd("   fastq_to_bam \"${prefix}\" $n");
+      script.cmd("   annotate_umi \"${prefix}\" $n");
+      script.cmd("done");
+      script.newLine();
+
       /*
-      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 toBamCmd = "./stdwrap.sh ./picard2 FastqToSam";
-      toBamCmd += " -SORT_ORDER queryname";
-      toBamCmd += " -QUALITY_FORMAT Standard";
-      toBamCmd += " ${rg_info}";
-      toBamCmd += " -FASTQ trim.quality/${prefix}_R1.fastq";
-      toBamCmd += " -FASTQ2 trim.quality/${prefix}_R2.fastq";
-      toBamCmd += " -OUTPUT bam/${prefix}.bam";
-      toBamCmd += " >> fastqtosam.out";
-      
-      String umiCmd = "./stdwrap.sh ${JAVA} -jar ${FgBioJAR} AnnotateBamWithUmis";
-      umiCmd += " --fail-fast true -t RX";
-      umiCmd += " -i bam/${prefix}.bam";
-      umiCmd += " -f fastq/${prefix}_UMI.fastq.gz";
-      umiCmd += " -o bam.umi/${prefix}.bam";
-      umiCmd += " >> annotatebam.out";
-      
-      script.cmd("   " + toBamCmd);
-      script.cmd("   " + umiCmd);
-      script.cmd("done");
-      script.newLine();
-    
       script.cmd("min_insert=$(awk 'NR == 1 || $3 - $2 < min {min = $3 - $2}END{print min - 1}' \"${AmpliconsBed}\")");
       script.cmd("max_insert=$(awk 'NR == 1 || $3 - $2 > max {max = $3 - $2}END{print max + 1}' \"${AmpliconsBed}\")");