Changeset 5815

Timestamp:

Jan 24, 2020, 2:37:45 PM (3 years ago)

Author:

Nicklas Nordborg

Message:

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

Updated the wizard to ask if the list contains tumor or normal samples. This option affect only the file name of the VCF that is cretated:

• Tumors: scanb-tumors.vcf.gz
• Normals: scanb-normals.vcf.gz

Location:

extensions/net.sf.basedb.reggie/trunk

Files:

• resources/analysis/vcall_build.js (modified) (1 diff)
• resources/analysis/vcall_build.jsp (modified) (1 diff)
• src/net/sf/basedb/reggie/grid/VariantStatisticsJobCreator.java (modified) (7 diffs)
• src/net/sf/basedb/reggie/servlet/VariantCallingServlet.java (modified) (2 diffs)

extensions/net.sf.basedb.reggie/trunk/resources/analysis/vcall_build.js

@@ -180 +180 @@
     submitInfo.itemList = parseInt(frm.itemList.value);
     submitInfo.cluster = frm.clusters.value;
+    submitInfo.sampleType = frm.sampleType.value;
     if (frm.priority.selectedIndex >= 0)
     {

extensions/net.sf.basedb.reggie/trunk/resources/analysis/vcall_build.jsp

@@ -67 +67 @@
           way up to a patient item. This is needed since the script calculates
           per-patient frequencies of variants.
+        </td>
+      </tr>
+      <tr>
+        <td class="subprompt">Sample type</td>
+        <td class="input">
+          <label><input type="radio" name="sampleType" id="sampleTypeTumors" value="tumors" checked>Tumors</label>
+          <label><input type="radio" name="sampleType" id="sampleTypeNormals" value="normals">Normals</label>
+        </td>
+        <td class="status" id="sampleType.status"></td>
+        <td class="help">
+          <span id="sampleType.message" class="message"></span>
+          Select the sample type of the alignments in the list. The selection
+          dosn't affect the analysis except the name of the final result files.
         </td>
       </tr>

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

@@ -21 +21 @@
 import net.sf.basedb.core.ProgressReporter;
 import net.sf.basedb.core.SessionControl;
+import net.sf.basedb.core.StringParameterType;
 import net.sf.basedb.core.Type;
 import net.sf.basedb.core.query.Expressions;
@@ -56 +57 @@
   private boolean debug;
   private Integer priority;
+  private String outputBaseName = "scanb-tumors";
   
   public VariantStatisticsJobCreator()
@@ -76 +78 @@
   {
     this.priority = priority;
+  }
+  
+  /**
+    Set the base name of files that are created.
+    Typically, this should be "scanb-tumors" or 
+    "scanb-normals" (.vcf.gz is and other extensions are automatically appended)
+  */
+  public void setOutputBaseName(String baseName)
+  {
+    this.outputBaseName = baseName;
   }
   
@@ -112 +124 @@
     statJob.setName("Build variant statistics");
     statJob.setParameterValue("list", new ItemParameterType<ItemList>(ItemList.class, null), list);
+    statJob.setParameterValue("outputBaseName", new StringParameterType(), outputBaseName);
     if (debug) statJob.setName(statJob.getName() + " (debug)");
     dc.saveItem(statJob);
@@ -118 +131 @@
     script.comment("Setting up scripting environment and copying script to tmp folder");
     script.cmd("export ScriptDir=" + pipeline_scripts_path);
+    script.cmd("OUTNAME="+outputBaseName);
     script.newLine();
 
@@ -129 +143 @@
     statCmd += " ${WD}/vcflist.txt";
     statCmd += " ${WD}/progress";
-    statCmd += " ${WD}/mut_stats.log";
-    statCmd += " > tmp/mut_stats.vcf";
+    statCmd += " ${WD}/${OUTNAME}.log";
+    statCmd += " > tmp/${OUTNAME}.vcf";
     script.cmd(statCmd);
     script.progress(95, "Sorting and indexing...");
-    script.cmd(bedtools_path+" sort -header -i tmp/mut_stats.vcf | bgzip -c > results/mut_stats.vcf.gz");
-    script.cmd("tabix results/mut_stats.vcf.gz");
+    script.cmd(bedtools_path+" sort -header -i tmp/${OUTNAME}.vcf | bgzip -c > results/${OUTNAME}.vcf.gz");
+    script.cmd("tabix results/${OUTNAME}.vcf.gz");
     script.cmd("cp results/* ${WD}");
     
@@ -214 +228 @@
     {
       String jobName = status.getName();
-      String log = session.getJobFileAsString(jobName, "mut_stats.log", "UTF-8");
+      String baseFileName = job.getParameterValue("outputBaseName");
+      String log = session.getJobFileAsString(jobName, baseFileName+".log", "UTF-8");
       Stats stat = Stats.parse(log);
       
       return "Parsed " + stat.numVcf + " VCF files for " + stat.numPatients + " patients." + 
         " Found "+Reggie.formatCount(stat.numVariants) + " variants." + 
-        " Result files can be found at: " + session.getHost().getWorkFolder(jobName);
+        " Result files can be found at: " + session.getHost().getWorkFolder(jobName) + 
+        "/" + baseFileName + ".*";
     }
   }

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

@@ -521 +521 @@
       
         Number listId = (Number)jsonReq.get("itemList");
+        boolean isNormals = "normals".equals(jsonReq.get("sampleType"));
         String clusterId = (String)jsonReq.get("cluster");
         boolean debug = Boolean.TRUE.equals(jsonReq.get("debug"));
@@ -539 +540 @@
         jobCreator.setDebug(debug);
         jobCreator.setPriority(priority == null ? null : priority.intValue());
+        jobCreator.setOutputBaseName(isNormals ? "scanb-normals" : "scanb-tumors");
 
         Job statJob = jobCreator.createVariantStatisticsJob(dc, cluster, list, progress);