Changeset 5795

Timestamp:

Dec 16, 2019, 2:33:10 PM (3 years ago)

Author:

Nicklas Nordborg

Message:

References #1215: Include mutation signature analysis in the variant calling pipeline

Implemented a step in the variant calling pipeline for calling the mutation_signature.R script.

Location:

extensions/net.sf.basedb.reggie/trunk

Files:

• config/reggie-config.xml (modified) (1 diff)
• src/net/sf/basedb/reggie/grid/VariantCallingJobCreator.java (modified) (4 diffs)

extensions/net.sf.basedb.reggie/trunk/config/reggie-config.xml

@@ -371 +371 @@
         <!-- static options for the SnpSift command (required) -->
         <snpsift-options>-s ${BaseDir}/rna_chr_set.txt -s ${BaseDir}/intogen-BRCA-genes-list_patch.txt -e ${BaseDir}/filter_expression.txt</snpsift-options>
+        
+        <!-- path to the COSMIC mutation signature data -->
+        <mutation-signature>${BaseDir}/COSMIC_Cancer_signatures_probabilities.RData</mutation-signature>
       </variant-call>
       

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

@@ -202 +202 @@
     String snpeff_options = cfg.getRequiredConfig("variant-call/snpeff-options", parameterSet);
     String snpsift_options = cfg.getRequiredConfig("variant-call/snpsift-options", parameterSet);
+    String cosmic_signatures_path = cfg.getRequiredConfig("variant-call/mutation-signature", parameterSet);
 
     // Selected items must be removed from this list
@@ -246 +247 @@
       
       String libName = ScriptUtil.checkValidScriptParameter(lib.getName());
-      // Remove all after the first dot not followed by a number
       if (specimen != null && specimen.getExternalId() != null)
       {
@@ -351 +351 @@
       script.cmd("cp ${ScriptDir}/gc_stat.toml .");
       script.cmd("cp ${ScriptDir}/g5_dist.toml .");
+      script.cmd("cp ${ScriptDir}/mutation_signature.R .");
       script.cmd("mkdir bam");
       script.cmd("mkdir mosdepth");
@@ -474 +475 @@
         script.cmd(snpSiftCmd);
         script.newLine();
+
+        // 'grep' to find number of variants The ' || true' is critical since 'grep' will exit with code 1 if there are no matches
+        script.cmd("NUMFILTERED=`grep -c -v '^#' resultsfilter/variants-filtered.vcf || true`");
+        script.cmd("if [ $NUMFILTERED -gt 0 ]; then");
+        String mutSignCmd = "./mutation_signature.R";
+        mutSignCmd += " " + libName;
+        mutSignCmd += " resultsfilter/variants-filtered.vcf";
+        mutSignCmd += " " + cosmic_signatures_path;
+        mutSignCmd += " resultsfilter/mutation_signature.pdf";
+        mutSignCmd += " > resultsfilter/mutation_signature.txt";
+        script.cmd(mutSignCmd);
+        script.newLine();
+        script.cmd("fi");
+        
         script.cmd("else");
         script.cmd("  cat tmp/variants-raw-2.vcf | bgzip -c > resultsfilter/variants-annotated.vcf.gz");