Changeset 6467

Timestamp:

Nov 3, 2021, 3:13:27 PM (15 months ago)

Author:

Nicklas Nordborg

Message:

References #1347: Updates to registration of specimen handled by external lab

Figure out which FASTQ is R1 and which is R2 by checking the stored annotation values instead of using the find command. The pattern used may not always match.

File:

• extensions/net.sf.basedb.reggie/branches/4.33-stable/src/net/sf/basedb/reggie/grid/ImportFastqJobCreator.java (modified) (4 diffs)

extensions/net.sf.basedb.reggie/branches/4.33-stable/src/net/sf/basedb/reggie/grid/ImportFastqJobCreator.java

@@ -187 +187 @@
             demuxName + " must have two values: " + rawFastqNames);
       }
+      String fastqNameR1 = null;
+      String fastqNameR2 = null;
+      // NOTE! Order in the list is not specified so we don't know which is R1 and R2
+      // Search for _R1 and _R2
+      for (String fqName : rawFastqNames)
+      {
+        if (fqName.contains("_R1")) fastqNameR1 = fqName;
+        if (fqName.contains("_R2")) fastqNameR2 = fqName;
+      }
+      if (fastqNameR1 == null)
+      {
+        throw new InvalidDataException("Annotation RawFASTQ on " + 
+            demuxName + " is missing an R1 fastq file: " + rawFastqNames);
+      }
+      if (fastqNameR2 == null)
+      {
+        throw new InvalidDataException("Annotation RawFASTQ on " + 
+            demuxName + " is missing an R2 fastq file: " + rawFastqNames);
+      }
       
       // Get SequencingRun so that we can get the path to the FASTQ folder.
@@ -272 +291 @@
       // If not, we need to copy from ImportGateway
       // If they don't exists at all we generate an error
-      int fileNo = 0;
       script.cmd("mkdir -p ${ImportArchive}");
-      for (String fastqName : rawFastqNames)
-      {
-        fileNo++;
-        // NOTE! Order in the list is not specified so we don't know which is R1 and R2
-        script.cmd("FASTQ="+ScriptUtil.checkValidFilename(fastqName));
-        script.progress(5+fileNo*5, "Copying FASTQ files: ${FASTQ}");
-        script.cmd("if [ ! -f \"${ImportArchive}/${FASTQ}\" ]; then");
-        script.cmd("  if [ ! -f \"${ImportGateway}/${FASTQ}\" ]; then");
-        script.cmd("    echo \"Can't find FASTQ file ${FASTQ} in ${ImportGateway} or ${ImportArchive}\" 1>&2");
+      script.cmd("FASTQ1="+ScriptUtil.checkValidFilename(fastqNameR1));
+      script.cmd("FASTQ2="+ScriptUtil.checkValidFilename(fastqNameR2));
+      for (int fileNo = 1; fileNo <= 2; fileNo++)
+      {
+        script.progress(5+fileNo*5, "Copying FASTQ files: ${FASTQ"+fileNo+"}");
+        script.cmd("if [ ! -f \"${ImportArchive}/${FASTQ"+fileNo+"}\" ]; then");
+        script.cmd("  if [ ! -f \"${ImportGateway}/${FASTQ"+fileNo+"}\" ]; then");
+        script.cmd("    echo \"Can't find FASTQ file ${FASTQ"+fileNo+"} in ${ImportGateway} or ${ImportArchive}\" 1>&2");
         script.cmd("    exit 1");
         script.cmd("   fi");
-        script.cmd("   cp \"${ImportGateway}/${FASTQ}\" \"${ImportArchive}/${FASTQ}\"");
+        script.cmd("   cp \"${ImportGateway}/${FASTQ"+fileNo+"}\" \"${ImportArchive}/${FASTQ"+fileNo+"}\"");
         if (!debug)
         {
-          script.cmd("   rm -f \"${ImportGateway}/${FASTQ}\"");
+          script.cmd("   rm -f \"${ImportGateway}/${FASTQ"+fileNo+"}\"");
         }
         script.cmd("fi");
-        script.cmd("cp \"${ImportArchive}/${FASTQ}\" fastq");
+        script.cmd("cp \"${ImportArchive}/${FASTQ"+fileNo+"}\" fastq");
         script.newLine();
       }
-      
-      script.comment("Find FASTQ files");
-      script.cmd("FASTQ1=`find fastq -name \"*_R1*.fastq.gz\" -print -quit 2> /dev/null`");
-      script.cmd("FASTQ2=`find fastq -name \"*_R2*.fastq.gz\" -print -quit 2> /dev/null`");
-      script.newLine();
 
       script.comment("Run Bowtie2");
-      script.progress(40, "Bowtie2: " + mergedName + " (${NumThreads} threads)");
+      script.progress(20, "Bowtie2: " + mergedName + " (${NumThreads} threads)");
       String alignCmd = "./stdwrap.sh " + bowtie_path;
       alignCmd += " -p ${NumThreads}";
@@ -307 +319 @@
       alignCmd += " --un-conc fastq.aligned/R%.fastq";
       alignCmd += " -x ${Gidx}";
-      alignCmd += " -1 ${FASTQ1}";
-      alignCmd += " -2 ${FASTQ2}";
+      alignCmd += " -1 fastq/${FASTQ1}";
+      alignCmd += " -2 fastq/${FASTQ2}";
       alignCmd += " -S fastq.aligned/aligned.sam";
       alignCmd += " > fastq.aligned/aligned.out";
@@ -319 +331 @@
       String trimCmd1 = "./stdwrap.sh ./trimmomatic PE";
       trimCmd1 += " -threads ${NumThreads}";
-      trimCmd1 += " ${FASTQ1}";
-      trimCmd1 += " ${FASTQ2}";
+      trimCmd1 += " fastq/${FASTQ1}";
+      trimCmd1 += " fastq/${FASTQ2}";
       trimCmd1 += " fastq.trimmomatic.1/"+R1_name;
       trimCmd1 += " fastq.trimmomatic.1/un_"+R1_name;