DEV-618: Remove manta/BPI

hartwigmedical · Dec 11, 2018 · 3ae8eae · 3ae8eae
1 parent 4698deb
commit 3ae8eae
Show file tree

Hide file tree

Showing 16 changed files with 84 additions and 243 deletions.
diff --git a/lib/HMF/Pipeline.pm b/lib/HMF/Pipeline.pm
@@ -16,7 +16,7 @@ use HMF::Pipeline::Amber;
 use HMF::Pipeline::Cobalt;
 use HMF::Pipeline::GermlineCalling;
 use HMF::Pipeline::Strelka;
-use HMF::Pipeline::StructuralVariants;
+use HMF::Pipeline::Gridss;
 use HMF::Pipeline::Purple;
 use HMF::Pipeline::HealthCheck;
 use HMF::Pipeline::PipelineCheck;
@@ -41,15 +41,15 @@ sub run {
     if (($opt->{FASTQ} and $opt->{MAPPING} eq "yes") or $opt->{BAM}) {
         # KODU: Always run post stats, even if we start from BAM.
         HMF::Pipeline::PostStats::run($opt) if $opt->{POSTSTATS} eq "yes" and $opt->{BAM};
+        HMF::Pipeline::DamageEstimate::run($opt) if $opt->{DAMAGE_ESTIMATE} eq "yes";
 
         HMF::Pipeline::Amber::run($opt) if $opt->{AMBER} eq "yes";
         HMF::Pipeline::Cobalt::run($opt) if $opt->{COBALT} eq "yes";
-        HMF::Pipeline::DamageEstimate::run($opt) if $opt->{DAMAGE_ESTIMATE} eq "yes";
 
         HMF::Pipeline::GermlineCalling::run($opt) if $opt->{GERMLINE_CALLING} eq "yes";
 
         HMF::Pipeline::Strelka::run($opt) if $opt->{STRELKA} eq "yes";
-        HMF::Pipeline::StructuralVariants::run($opt) if $opt->{STRUCTURAL_VARIANT_CALLING} eq "yes";
+        HMF::Pipeline::Gridss::run($opt) if $opt->{GRIDSS} eq "yes";
         HMF::Pipeline::Purple::run($opt) if $opt->{PURPLE} eq "yes";
 
         HMF::Pipeline::HealthCheck::run($opt) if $opt->{HEALTHCHECK} eq "yes";

diff --git a/lib/HMF/Pipeline/Functions/Validate.pm b/lib/HMF/Pipeline/Functions/Validate.pm
@@ -355,6 +355,9 @@ sub configChecks {
                 PURPLE_TIME => \&key_not_present,
                 PURPLE_THREADS => \&key_not_present,
                 PURPLE_MEM => \&key_not_present,
+
+                PURPLE_REF_GENOME_VERSION => \&key_not_present,
+
                 CIRCOS_PATH => \&missing_directory,
                 GC_PROFILE => \&missing_file,
             }
@@ -406,63 +409,42 @@ sub configChecks {
                 TABIX_PATH => \&missing_directory,
             }
         ),
-        STRUCTURAL_VARIANT_CALLING => if_enabled({
-                MANTA => if_enabled({
-                        MANTA_PATH => \&missing_directory,
-                        MANTA_QUEUE => \&key_not_present,
-                        MANTA_THREADS => \&key_not_present,
-                        MANTA_MEM => \&key_not_present,
-                        MANTA_TIME => \&key_not_present,
-
-                        BPI_PATH => \&missing_directory,
-                        BPI_QUEUE => \&key_not_present,
-                        BPI_THREADS => \&key_not_present,
-                        BPI_MEM => \&key_not_present,
-                        BPI_TIME => \&key_not_present,
-
-                        BPI_CONTAMINATION_FRACTION => \&key_not_present,
-
-                        TABIX_PATH => \&missing_directory,
-                    }
-                ),
-                GRIDSS => if_enabled({
-                        GRIDSS_PATH => \&missing_directory,
-                        GRIDSS_BWA_PATH => \&missing_directory,
-
-                        GRIDSS_PREPROCESS_QUEUE => \&key_not_present,
-                        GRIDSS_PREPROCESS_THREADS => \&key_not_present,
-                        GRIDSS_PREPROCESS_MEM => \&key_not_present,
-                        GRIDSS_PREPROCESS_TIME => \&key_not_present,
-
-                        GRIDSS_ASSEMBLE_QUEUE => \&key_not_present,
-                        GRIDSS_ASSEMBLE_THREADS => \&key_not_present,
-                        GRIDSS_ASSEMBLE_MEM => \&key_not_present,
-                        GRIDSS_ASSEMBLE_TIME => \&key_not_present,
-
-                        GRIDSS_ASSEMBLE_POST_PROCESS_QUEUE => \&key_not_present,
-                        GRIDSS_ASSEMBLE_POST_PROCESS_THREADS => \&key_not_present,
-                        GRIDSS_ASSEMBLE_POST_PROCESS_MEM => \&key_not_present,
-                        GRIDSS_ASSEMBLE_POST_PROCESS_TIME => \&key_not_present,
-
-                        GRIDSS_CALLING_QUEUE => \&key_not_present,
-                        GRIDSS_CALLING_THREADS => \&key_not_present,
-                        GRIDSS_CALLING_MEM => \&key_not_present,
-                        GRIDSS_CALLING_TIME => \&key_not_present,
-
-                        GRIDSS_ANNOTATE_QUEUE => \&key_not_present,
-                        GRIDSS_ANNOTATE_THREADS => \&key_not_present,
-                        GRIDSS_ANNOTATE_MEM => \&key_not_present,
-                        GRIDSS_ANNOTATE_TIME => \&key_not_present,
-
-                        GRIDSS_THRESHOLD_COVERAGE => \&key_not_present,
-                        GRIDSS_BWA_BASES_PER_BATCH => \&key_not_present,
-                        GRIDSS_BLACKLIST => \&missing_file,
-                        GRIDSS_CONFIG => \&key_not_present,
-                        GRIDSS_PON => \&missing_directory,
-
-                        TABIX_PATH => \&missing_directory,
-                    }
-                ),
+        GRIDSS => if_enabled({
+                GRIDSS_PATH => \&missing_directory,
+                GRIDSS_BWA_PATH => \&missing_directory,
+
+                GRIDSS_PREPROCESS_QUEUE => \&key_not_present,
+                GRIDSS_PREPROCESS_THREADS => \&key_not_present,
+                GRIDSS_PREPROCESS_MEM => \&key_not_present,
+                GRIDSS_PREPROCESS_TIME => \&key_not_present,
+
+                GRIDSS_ASSEMBLE_QUEUE => \&key_not_present,
+                GRIDSS_ASSEMBLE_THREADS => \&key_not_present,
+                GRIDSS_ASSEMBLE_MEM => \&key_not_present,
+                GRIDSS_ASSEMBLE_TIME => \&key_not_present,
+
+                GRIDSS_ASSEMBLE_POST_PROCESS_QUEUE => \&key_not_present,
+                GRIDSS_ASSEMBLE_POST_PROCESS_THREADS => \&key_not_present,
+                GRIDSS_ASSEMBLE_POST_PROCESS_MEM => \&key_not_present,
+                GRIDSS_ASSEMBLE_POST_PROCESS_TIME => \&key_not_present,
+
+                GRIDSS_CALLING_QUEUE => \&key_not_present,
+                GRIDSS_CALLING_THREADS => \&key_not_present,
+                GRIDSS_CALLING_MEM => \&key_not_present,
+                GRIDSS_CALLING_TIME => \&key_not_present,
+
+                GRIDSS_ANNOTATE_QUEUE => \&key_not_present,
+                GRIDSS_ANNOTATE_THREADS => \&key_not_present,
+                GRIDSS_ANNOTATE_MEM => \&key_not_present,
+                GRIDSS_ANNOTATE_TIME => \&key_not_present,
+
+                GRIDSS_THRESHOLD_COVERAGE => \&key_not_present,
+                GRIDSS_BWA_BASES_PER_BATCH => \&key_not_present,
+                GRIDSS_BLACKLIST => \&missing_file,
+                GRIDSS_CONFIG => \&key_not_present,
+                GRIDSS_PON => \&missing_directory,
+
+                TABIX_PATH => \&missing_directory,
             }
         ),
         HEALTHCHECK => if_enabled({

diff --git a/lib/HMF/Pipeline/StructuralVariants.pm → lib/HMF/Pipeline/Gridss.pm b/lib/HMF/Pipeline/StructuralVariants.pm → lib/HMF/Pipeline/Gridss.pm
@@ -1,4 +1,4 @@
-package HMF::Pipeline::StructuralVariants;
+package HMF::Pipeline::Gridss;
 
 use FindBin::libs;
 use discipline;
@@ -20,36 +20,30 @@ sub run {
 
     say "\n### SCHEDULING STRUCTURAL VARIANT CALLING ###";
 
-    $opt->{RUNNING_JOBS}->{sv} = [];
-    if ($opt->{MANTA} eq "yes") {
-        my $manta_jobs = runManta($opt);
-        push @{$opt->{RUNNING_JOBS}->{sv}}, @{$manta_jobs};
-    }
+    $opt->{RUNNING_JOBS}->{gridss} = [];
+
+    if ($opt->{POSTSTATS} eq "no") {
+        say "\n[ERROR] Cannot schedule gridss without scheduling post stats!";
+    } else {
+        # KODU: We need the insert size metrics when running gridss. Their naming comes out of poststats and is dependent on the mode we run in.
+        my ($ref_sample, $tumor_sample, undef, undef, undef, undef) = sampleControlBamsAndJobs($opt);
+        my $suffix = "_MultipleMetrics.txt.insert_size_metrics";
+        my $ref_sample_name;
+        my $tumor_sample_name;
+        if ($opt->{BAM}) {
+            $ref_sample_name = $ref_sample;
+            $tumor_sample_name = $tumor_sample;
+        } elsif ($opt->{FASTQ}) {
+            $ref_sample_name = join "", $ref_sample, "_dedup";
+            $tumor_sample_name = join "", $tumor_sample, "_dedup";
+        }
+
+        $opt->{REF_INSERT_SIZE_METRICS} = catfile($opt->{OUTPUT_DIR}, "QCStats", $ref_sample_name, join "", $ref_sample_name, $suffix);
+        $opt->{TUMOR_INSERT_SIZE_METRICS} = catfile($opt->{OUTPUT_DIR}, "QCStats", $tumor_sample_name, join "", $tumor_sample_name, $suffix);
 
-    if ($opt->{GRIDSS} eq "yes") {
-        if ($opt->{POSTSTATS} eq "no") {
-            say "\n[WARN] Cannot schedule gridss without scheduling post stats!";
-        } else {
-            # KODU: We need the insert size metrics when running gridss. Their naming comes out of poststats and is dependent on the mode we run in.
-            my ($ref_sample, $tumor_sample, undef, undef, undef, undef) = sampleControlBamsAndJobs($opt);
-            my $suffix = "_MultipleMetrics.txt.insert_size_metrics";
-            my $ref_sample_name;
-            my $tumor_sample_name;
-            if ($opt->{BAM}) {
-                $ref_sample_name = $ref_sample;
-                $tumor_sample_name = $tumor_sample;
-            } elsif ($opt->{FASTQ}) {
-                $ref_sample_name = join "", $ref_sample, "_dedup";
-                $tumor_sample_name = join "", $tumor_sample, "_dedup";
-            }
-
-            $opt->{REF_INSERT_SIZE_METRICS} = catfile($opt->{OUTPUT_DIR}, "QCStats", $ref_sample_name, join "", $ref_sample_name, $suffix);
-            $opt->{TUMOR_INSERT_SIZE_METRICS} = catfile($opt->{OUTPUT_DIR}, "QCStats", $tumor_sample_name, join "", $tumor_sample_name, $suffix);
-
-            my $gridss_jobs = runGridss($opt);
-            if ($gridss_jobs) {
-                push @{$opt->{RUNNING_JOBS}->{sv}}, @{$gridss_jobs};
-            }
+        my $gridss_jobs = runGridss($opt);
+        if ($gridss_jobs) {
+            push @{$opt->{RUNNING_JOBS}->{gridss}}, @{$gridss_jobs};
         }
     }
 
@@ -256,6 +250,8 @@ 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");
+
     # KODU: Run with GRIDSS annotate settings, filter takes little resources.
     my $job_id = fromTemplate(
         "GridssFilter",
@@ -267,76 +263,14 @@ sub runGridssFilter {
         $opt,
         tumor_sample => $tumor_sample,
         joint_name => $joint_name,
+        final_vcf => ${final_vcf},
     );
 
-    return ($job_id);
-}
-
-sub runManta {
-    my ($opt) = @_;
-
-    say "\n### SCHEDULING MANTA ###";
-
-    my ($ref_sample, $tumor_sample, $ref_sample_bam, $tumor_sample_bam, $joint_name, $running_jobs) = sampleControlBamsAndJobs($opt);
-
-    my @manta_jobs;
-    my $job_id = runMantaJob($tumor_sample_bam, $ref_sample_bam, $joint_name, $running_jobs, $opt);
-    push @manta_jobs, $job_id;
-
-    $job_id = runBreakpointInspector($tumor_sample, $tumor_sample_bam, $ref_sample, $ref_sample_bam, $joint_name, \@manta_jobs, $opt);
-    push @manta_jobs, $job_id;
-    return \@manta_jobs;
-}
-
-sub runMantaJob {
-    my ($tumor_sample_bam, $ref_sample_bam, $joint_name, $running_jobs, $opt) = @_;
-
-    my $dirs = createDirs(catfile($opt->{OUTPUT_DIR}, "structuralVariants", "manta", $joint_name));
-
-    my $job_id = fromTemplate(
-        "Manta",
-        undef,
-        1,
-        qsubTemplate($opt, "MANTA"),
-        $running_jobs,
-        $dirs,
-        $opt,
-        ref_sample_bam => $ref_sample_bam,
-        tumor_sample_bam => $tumor_sample_bam,
-        joint_name => $joint_name,
-    );
-
-    return $job_id;
-}
-
-sub runBreakpointInspector {
-    my ($tumor_sample, $tumor_sample_bam, $ref_sample, $ref_sample_bam, $joint_name, $dependent_job_ids, $opt) = @_;
-
-    my $manta_vcf = catfile($opt->{OUTPUT_DIR}, "structuralVariants", "manta", $joint_name, "results", "variants", "somaticSV.vcf.gz");
-
-    my $dirs = createDirs(catfile($opt->{OUTPUT_DIR}, "structuralVariants", "bpi", $joint_name));
-    $opt->{STRUCTURAL_VARIANT_VCF} = catfile($dirs->{out}, "${joint_name}_somaticSV_bpi.vcf");
-
-    my $job_id = fromTemplate(
-        "BreakpointInspector",
-        undef,
-        1,
-        qsubTemplate($opt, "BPI"),
-        $dependent_job_ids,
-        $dirs,
-        $opt,
-        ref_sample => $ref_sample,
-        tumor_sample => $tumor_sample,
-        ref_sample_bam => $ref_sample_bam,
-        tumor_sample_bam => $tumor_sample_bam,
-        joint_name => $joint_name,
-        input_vcf => $manta_vcf,
-    );
-
-    $opt->{STRUCTURAL_VARIANT_VCF} = join "", $opt->{STRUCTURAL_VARIANT_VCF}, ".gz";
+    $opt->{STRUCTURAL_VARIANT_VCF} = $final_vcf;
+    $opt->{GRIDSS_VCF} = $final_vcf;
     linkVcfArtefacts($opt->{STRUCTURAL_VARIANT_VCF}, 'structural_variant', $opt);
 
-    return $job_id;
+    return ($job_id);
 }
 
 1;
diff --git a/lib/HMF/Pipeline/Purple.pm b/lib/HMF/Pipeline/Purple.pm
@@ -59,7 +59,7 @@ sub dependencies {
     my @jobs;
     push @jobs, @{$opt->{RUNNING_JOBS}->{amber}};
     push @jobs, @{$opt->{RUNNING_JOBS}->{cobalt}};
-    push @jobs, @{$opt->{RUNNING_JOBS}->{sv}};
+    push @jobs, @{$opt->{RUNNING_JOBS}->{gridss}};
     push @jobs, @{$opt->{RUNNING_JOBS}->{strelka}};
     return \@jobs;
 }

diff --git a/settings/CheckCoverage.ini b/settings/CheckCoverage.ini
@@ -11,7 +11,6 @@ STRELKA	no
 AMBER	no
 COBALT	no
 PURPLE	no
-
-STRUCTURAL_VARIANT_CALLING	no
+GRIDSS	no
 
 HEALTHCHECK	no
diff --git a/settings/PurpleOnly.ini b/settings/PurpleOnly.ini
@@ -11,10 +11,6 @@ STRELKA	yes
 AMBER	yes
 COBALT	yes
 PURPLE	yes
-
-STRUCTURAL_VARIANT_CALLING	yes
-MANTA	yes
-
-GRIDSS	no
+GRIDSS	yes
 
 HEALTHCHECK	no
diff --git a/settings/SingleSample.ini b/settings/SingleSample.ini
@@ -11,7 +11,6 @@ STRELKA	no
 AMBER	no
 COBALT	no
 PURPLE	no
-
-STRUCTURAL_VARIANT_CALLING	no
+GRIDSS	no
 
 HEALTHCHECK	yes
diff --git a/settings/Somatic.ini b/settings/Somatic.ini
@@ -11,9 +11,6 @@ STRELKA	yes
 AMBER	yes
 COBALT	yes
 PURPLE	yes
-
-STRUCTURAL_VARIANT_CALLING	yes
-MANTA	yes
 GRIDSS	yes
 
 HEALTHCHECK	yes
diff --git a/settings/include/cluster.ini b/settings/include/cluster.ini
@@ -101,16 +101,6 @@ COBALT_TIME	256:0:0
 COBALT_THREADS	6
 COBALT_MEM	10
 
-MANTA_QUEUE	all.q
-MANTA_TIME	256:0:0
-MANTA_THREADS	8
-MANTA_MEM	10
-
-BPI_QUEUE	all.q
-BPI_TIME	256:0:0
-BPI_THREADS	1
-BPI_MEM	4
-
 GRIDSS_PREPROCESS_QUEUE	all.q
 GRIDSS_PREPROCESS_TIME	256:0:0
 GRIDSS_PREPROCESS_THREADS	2

diff --git a/settings/include/settings.ini b/settings/include/settings.ini
@@ -29,7 +29,7 @@ ANNOTATE_SNPEFF_FLAGS	-hgvs -lof -no-downstream -ud 1000 -no-intergenic -noShift
 ANNOTATE_DBNSFP_FIELDS	hg38_chr,hg38_pos,genename,Uniprot_acc,Uniprot_id,Uniprot_aapos,Interpro_domain,cds_strand,refcodon,SLR_test_statistic,codonpos,fold-degenerate,Ancestral_allele,Ensembl_geneid,Ensembl_transcriptid,aapos,aapos_SIFT,aapos_FATHMM,SIFT_score,SIFT_converted_rankscore,SIFT_pred,Polyphen2_HDIV_score,Polyphen2_HDIV_rankscore,Polyphen2_HDIV_pred,Polyphen2_HVAR_score,Polyphen2_HVAR_rankscore,Polyphen2_HVAR_pred,LRT_score,LRT_converted_rankscore,LRT_pred,MutationTaster_score,MutationTaster_converted_rankscore,MutationTaster_pred,MutationAssessor_score,MutationAssessor_rankscore,MutationAssessor_pred,FATHMM_score,FATHMM_rankscore,FATHMM_pred,MetaSVM_score,MetaSVM_rankscore,MetaSVM_pred,MetaLR_score,MetaLR_rankscore,MetaLR_pred,Reliability_index,VEST3_score,VEST3_rankscore,PROVEAN_score,PROVEAN_converted_rankscore,PROVEAN_pred,CADD_raw,CADD_raw_rankscore,CADD_phred,GERP++_NR,GERP++_RS,GERP++_RS_rankscore,phyloP46way_primate,phyloP46way_primate_rankscore,phyloP46way_placental,phyloP46way_placental_rankscore,phyloP100way_vertebrate,phyloP100way_vertebrate_rankscore,phastCons46way_primate,phastCons46way_primate_rankscore,phastCons46way_placental,phastCons46way_placental_rankscore,phastCons100way_vertebrate,phastCons100way_vertebrate_rankscore,SiPhy_29way_pi,SiPhy_29way_logOdds,SiPhy_29way_logOdds_rankscore,LRT_Omega,UniSNP_ids,1000Gp1_AC,1000Gp1_AF,1000Gp1_AFR_AC,1000Gp1_AFR_AF,1000Gp1_EUR_AC,1000Gp1_EUR_AF,1000Gp1_AMR_AC,1000Gp1_AMR_AF,1000Gp1_ASN_AC,1000Gp1_ASN_AF,ESP6500_AA_AF,ESP6500_EA_AF,ARIC5606_AA_AC,ARIC5606_AA_AF,ARIC5606_EA_AC,ARIC5606_EA_AF,ExAC_AC,ExAC_AF,ExAC_Adj_AC,ExAC_Adj_AF,ExAC_AFR_AC,ExAC_AFR_AF,ExAC_AMR_AC,ExAC_AMR_AF,ExAC_EAS_AC,ExAC_EAS_AF,ExAC_FIN_AC,ExAC_FIN_AF,ExAC_NFE_AC,ExAC_NFE_AF,ExAC_SAS_AC,ExAC_SAS_AF,clinvar_rs,clinvar_clnsig,clinvar_trait,COSMIC_ID,COSMIC_CNT
 ANNOTATE_FREQ_INFO	AF,AN,AC
 
-BPI_CONTAMINATION_FRACTION	0.03
+PURPLE_REF_GENOME_VERSION	hg19
 
 GRIDSS_THRESHOLD_COVERAGE	50000
 GRIDSS_BWA_BASES_PER_BATCH	40000000
diff --git a/settings/include/tools.ini b/settings/include/tools.ini
@@ -19,8 +19,6 @@ COBALT_PATH	/data/tools/cobalt_v1.4
 PURPLE_PATH	/data/tools/purple_v2.17
 CIRCOS_PATH	/data/tools/circos_v0.69.6/bin
 
-MANTA_PATH	/data/tools/manta_v1.0.3/bin
-BPI_PATH	/data/tools/break-point-inspector_v1.7
 GRIDSS_PATH	/data/tools/gridss_v2.0.1
 GRIDSS_BWA_PATH	/data/tools/bwa_v0.7.17