Changeset 5768

Timestamp:

Dec 2, 2019, 2:22:53 PM (4 years ago)

Author:

Nicklas Nordborg

Message:

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

Started to implement the script for calculating the variant statistics. Most of the work is done by the mut_stats.py python script. After that we need to sort and index the result. The final VCF file is currently left in the working directory which is obviously not a good place.

File:

• extensions/net.sf.basedb.reggie/trunk/src/net/sf/basedb/reggie/grid/VariantStatisticsJobCreator.java (modified) (3 diffs)

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

@@ -91 +91 @@
     }
     
+    String projectRoot = cfg.getRequiredConfig("project-archive", null);
+    String pipeline_scripts_path = cfg.getRequiredConfig("programs/pipeline-scripts/path", null);
+    String bedtools_path = cfg.getRequiredConfig("programs/bedtools/path", null);
+
     // Options for the job
     JobConfig jobConfig = new JobConfig();
@@ -110 +114 @@
     ScriptBuilder script = new ScriptBuilder();
     script.comment("Setting up scripting environment and copying script to tmp folder");
+    script.cmd("export ScriptDir=" + pipeline_scripts_path);
+    script.newLine();
+
+    script.cmd("cd ${TMPDIR}");
+    script.cmd("cp ${ScriptDir}/mut_stats.py .");
+    script.newLine();
     
-    // TODO -- Create script here
+    String statCmd = "./mut_stats.py";
+    statCmd += " ${WD}/vcflist.txt";
+    statCmd += " > mut_stats.vcf";
+    script.cmd(statCmd);
+    script.cmd(bedtools_path+" sort -header -i mut_stats.vcf | bgzip -c > mut_stats.vcf.gz");
+    script.cmd("tabix mut_stats.vcf.gz");
     
     script.progress(99, "Cleaning up temporary folders");
@@ -163 +178 @@
       }
       String patient = patients.get(0).getName();
-      String dataFolder = (String)Annotationtype.DATA_FILES_FOLDER.getAnnotationValue(dc, manager, alignment);
-      tw.tablePrintData(patient, dataFolder+"/variants-raw.vcf.gz");
+      String dataFolder = ScriptUtil.checkValidPath((String)Annotationtype.DATA_FILES_FOLDER.getAnnotationValue(dc, manager, alignment), true, true);
+      tw.tablePrintData(patient, projectRoot + dataFolder+"/variants-raw.vcf.gz");
     }
     tw.flush();