Changeset 5772

Timestamp:

Dec 3, 2019, 11:56:24 AM (4 years ago)

Author:

Nicklas Nordborg

Message:

References #1208: Implement wizard for building database of variant frequencies in SCAN-B samples

Added a new job subtype so that we can get a callback when the job has ended. The callback will get some information from the log file and generate a message.

Location:

extensions/net.sf.basedb.reggie/trunk/src/net/sf/basedb/reggie

Files:

• dao/Subtype.java (modified) (1 diff)
• grid/JobCompletionHandlerFactory.java (modified) (1 diff)
• grid/VariantStatisticsJobCreator.java (modified) (4 diffs)
• servlet/InstallServlet.java (modified) (1 diff)

extensions/net.sf.basedb.reggie/trunk/src/net/sf/basedb/reggie/dao/Subtype.java

@@ -324 +324 @@
   public static final Subtype VARIANT_CALLING_JOB = new Subtype("Variant calling", null, null, null, Item.JOB, false);
 
+  /**
+    The definition of the variant calling job.
+    @since 4.25
+  */
+  public static final Subtype VARIANT_STATISTICS_JOB = new Subtype("Variant statistics", null, null, null, Item.JOB, false);
+  
   /**
     The definition of the "Sequencer" hardware subtype. Should already exist in BASE.

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

@@ -103 +103 @@
           action = new VariantCallingJobCreator.VariantCallJobCompletionHandler();
         }
+        else if (jobType.equals(Subtype.VARIANT_STATISTICS_JOB.get(dc)))
+        {
+          action = new VariantStatisticsJobCreator.VariantStatisticsJobCompletionHandler();
+        }
       }
       finally

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

@@ -27 +27 @@
 import net.sf.basedb.core.snapshot.SnapshotManager;
 import net.sf.basedb.opengrid.JobDefinition;
+import net.sf.basedb.opengrid.JobStatus;
 import net.sf.basedb.opengrid.OpenGridCluster;
+import net.sf.basedb.opengrid.OpenGridSession;
 import net.sf.basedb.opengrid.ScriptBuilder;
 import net.sf.basedb.opengrid.config.ClusterConfig;
 import net.sf.basedb.opengrid.config.JobConfig;
 import net.sf.basedb.opengrid.filetransfer.StringUploadSource;
+import net.sf.basedb.opengrid.service.JobCompletionHandler;
 import net.sf.basedb.reggie.Reggie;
 import net.sf.basedb.reggie.XmlConfig;
@@ -104 +107 @@
     // Create job 
     Job statJob = Job.getNew(dc, null, null, null);
-//    statJob.setItemSubtype(Subtype.VARIANT_CALLING_JOB.get(dc));
+    statJob.setItemSubtype(Subtype.VARIANT_STATISTICS_JOB.get(dc));
     statJob.setPluginVersion("reggie-"+Reggie.VERSION);
     statJob.setSendMessage(Values.getBoolean(sc.getUserClientSetting("plugins.sendmessage"), false));
@@ -124 +127 @@
     statCmd += " ${WD}/vcflist.txt";
     statCmd += " ${WD}/progress";
+    statCmd += " ${WD}/mut_stats.log";
     statCmd += " > mut_stats.vcf";
     script.cmd(statCmd);
+    script.progress(95, "Sorting and indexing...");
     script.cmd(bedtools_path+" sort -header -i mut_stats.vcf | bgzip -c > mut_stats.vcf.gz");
     script.cmd("tabix mut_stats.vcf.gz");
@@ -195 +200 @@
   }
 
-  
+  public static class VariantStatisticsJobCompletionHandler
+    implements JobCompletionHandler
+  {
+    
+    public VariantStatisticsJobCompletionHandler()
+    {}
+  
+    @Override
+    public String jobCompleted(SessionControl sc, OpenGridSession session, Job job, JobStatus status)
+    {
+      String jobName = status.getName();
+      String log = session.getJobFileAsString(jobName, "mut_stats.log", "UTF-8");
+      Stats stat = Stats.parse(log);
+      
+      return "Parsed " + stat.numVcf + " VCF files for " + stat.numPatients + " patients." + 
+        " Found "+Reggie.formatCount(stat.numVariants) + " variants.";
+    }
+  }
+
+  static class Stats
+  {
+    int numVcf = 0;
+    int numPatients = 0;
+    int numVariants = 0;
+    
+    static Stats parse(String statsOut)
+    {
+      Stats s = new Stats();
+      for (String line : statsOut.split("\n"))
+      {
+        String[] stat = line.split(":", 2);
+        String key = stat[0].strip();
+        int val = Values.getInt(stat[1].strip());
+        if ("NumberOfVCF".equals(key))
+        {
+          s.numVcf = val;
+        }
+        else if ("NumberOfPatients".equals(key))
+        {
+          s.numPatients = val;
+        }
+        else if ("NumberOfVariants".equals(key))
+        {
+          s.numVariants = val;
+        }
+      }
+      return s;
+    }
+  }
 }

extensions/net.sf.basedb.reggie/trunk/src/net/sf/basedb/reggie/servlet/InstallServlet.java

@@ -332 +332 @@
         jsonChecks.add(checkSubtype(dc, Subtype.MBAF_JOB, null, createIfMissing));
         jsonChecks.add(checkSubtype(dc, Subtype.VARIANT_CALLING_JOB, null, createIfMissing));
+        jsonChecks.add(checkSubtype(dc, Subtype.VARIANT_STATISTICS_JOB, null, createIfMissing));
         jsonChecks.add(checkSubtype(dc, Subtype.REPORT_SOFTWARE, null, createIfMissing));
         jsonChecks.add(checkSubtype(dc, Subtype.REPORT_JOB, null, createIfMissing));