DEV-618: Let purple generate final SV vcf

hartwigmedical · Dec 11, 2018 · 96150ef · 96150ef
1 parent 86401b2
commit 96150ef
Show file tree

Hide file tree

Showing 4 changed files with 23 additions and 19 deletions.
diff --git a/lib/HMF/Pipeline/Gridss.pm b/lib/HMF/Pipeline/Gridss.pm
@@ -10,7 +10,6 @@ use List::MoreUtils qw(uniq);
 use HMF::Pipeline::Functions::Config qw(createDirs sampleControlBamsAndJobs);
 use HMF::Pipeline::Functions::Job qw(fromTemplate checkReportedDoneFile markDone);
 use HMF::Pipeline::Functions::Sge qw(qsubTemplate);
-use HMF::Pipeline::Functions::Metadata qw(linkVcfArtefacts);
 
 use parent qw(Exporter);
 our @EXPORT_OK = qw(run);
@@ -250,7 +249,8 @@ sub runGridssCleanup {
 sub runGridssFilter {
     my ($dirs, $tumor_sample, $joint_name, $dependent_jobs, $opt) = @_;
 
-    my $final_vcf = catfile($dirs->{out}, join "", ${tumor_sample}, ".gridss.somatic.vcf.gz");
+    my $filtered_vcf = catfile($dirs->{out}, join "", ${tumor_sample}, ".gridss.somatic.vcf.gz");
+    my $full_vcf = catfile($dirs->{out}, join "", ${tumor_sample}, ".gridss.somatic.full.vcf.gz");
 
     # KODU: Run with GRIDSS annotate settings, filter takes little resources.
     my $job_id = fromTemplate(
@@ -263,12 +263,12 @@ sub runGridssFilter {
         $opt,
         tumor_sample => $tumor_sample,
         joint_name => $joint_name,
-        final_vcf => ${final_vcf},
+        filtered_vcf => $filtered_vcf,
+        full_vcf => $full_vcf,
     );
 
-    $opt->{STRUCTURAL_VARIANT_VCF} = $final_vcf;
-    $opt->{GRIDSS_VCF} = $final_vcf;
-    linkVcfArtefacts($opt->{STRUCTURAL_VARIANT_VCF}, 'structural_variant', $opt);
+    $opt->{GRIDSS_FILTERED_VCF} = $filtered_vcf;
+    $opt->{GRIDSS_FULL_VCF} = $full_vcf;
 
     return ($job_id);
 }

diff --git a/lib/HMF/Pipeline/Purple.pm b/lib/HMF/Pipeline/Purple.pm
@@ -8,7 +8,7 @@ use File::Spec::Functions;
 use HMF::Pipeline::Functions::Config qw(createDirs);
 use HMF::Pipeline::Functions::Sge qw(qsubJava);
 use HMF::Pipeline::Functions::Job qw(fromTemplate);
-use HMF::Pipeline::Functions::Metadata qw(parse linkArtefact);
+use HMF::Pipeline::Functions::Metadata qw(parse linkArtefact linkVcfArtefacts);
 
 use parent qw(Exporter);
 our @EXPORT_OK = qw(run);
@@ -20,15 +20,11 @@ sub run {
 
     say "\n### SCHEDULING PURPLE ###";
 
-    my $sub_dir = "purple";
-    my $dirs = createDirs($opt->{OUTPUT_DIR}, purple => $sub_dir);
+    my $purple_dir = "purple";
+    my $dirs = createDirs($opt->{OUTPUT_DIR}, purple => $purple_dir);
     my $dependent_jobs = dependencies($opt);
 
-    my $circos_path = "${sub_dir}/plot/${tumor_sample}.circos.png";
-    my $purple_cnv = "${sub_dir}/${tumor_sample}.purple.cnv";
-    my $purple_gene_cnv = "${sub_dir}/${tumor_sample}.purple.gene.cnv";
-    my $purple_germline_cnv = "${sub_dir}/${tumor_sample}.purple.germline.cnv";
-    my $purple_purity = "${sub_dir}/${tumor_sample}.purple.purity";
+    my $purple_purity = "$purple_dir/${tumor_sample}.purple.purity";
 
     my $job_id = fromTemplate(
         "Purple",
@@ -44,6 +40,13 @@ sub run {
 
     $opt->{RUNNING_JOBS}->{purple} = [$job_id] if $job_id;
 
+    my $purple_sv_path = "${purple_dir}/${tumor_sample}.purple.sv.vcf.gz";
+    my $circos_path = "${purple_dir}/plot/${tumor_sample}.circos.png";
+    my $purple_cnv = "${purple_dir}/${tumor_sample}.purple.cnv";
+    my $purple_gene_cnv = "${purple_dir}/${tumor_sample}.purple.gene.cnv";
+    my $purple_germline_cnv = "${purple_dir}/${tumor_sample}.purple.germline.cnv";
+
+    linkVcfArtefacts($purple_sv_path, 'structural_variant', $opt);
     linkArtefact($circos_path, 'circos_plot', $opt);
     linkArtefact($purple_cnv, 'purple_cnv', $opt);
     linkArtefact($purple_gene_cnv, 'purple_gene_cnv', $opt);

diff --git a/templates/GridssFilter.sh.tt b/templates/GridssFilter.sh.tt
@@ -20,10 +20,10 @@ Rscript ${libgridss}/gridss_somatic_filter.R -p ${gridss_pon} -i ${unzipped_orig
 
 rm ${unzipped_original_vcf}
 
-mv ${full_vcf}.bgz ${full_vcf}.gz
-mv ${full_vcf}.bgz.tbi ${full_vcf}.gz.tbi
+mv ${full_vcf}.bgz [% full_vcf %]
+mv ${full_vcf}.bgz.tbi [% full_vcf %].tbi
 
-mv ${output_vcf}.bgz [% final_vcf %]
-mv ${output_vcf}.bgz.tbi [% final_vcf %].tbi
+mv ${output_vcf}.bgz [% filtered_vcf %]
+mv ${output_vcf}.bgz.tbi [% filtered_vcf %].tbi
 
 success
diff --git a/templates/Purple.sh.tt b/templates/Purple.sh.tt
@@ -10,7 +10,8 @@ java -Xmx[% opt.PURPLE_MEM %]G -jar "[% opt.PURPLE_PATH %]/purple.jar" \
     -threads [% opt.PURPLE_THREADS %] \
     -gc_profile "[% opt.GC_PROFILE %]" \
     -somatic_vcf [% opt.SOMATIC_VARIANT_VCF %]  \
-    -structural_vcf [% opt.GRIDSS_VCF %] \
+    -structural_vcf [% opt.GRIDSS_FILTERED_VCF %] \
+    -sv_recovery_vcf [% opt.GRIDSS_FULL_VCF %] \
     -baf [% opt.AMBER_BAF_FILE %] \
     -circos [% opt.CIRCOS_PATH %]/circos \
     -ref_genome [%  opt.PURPLE_REF_GENOME_VERSION %]  \