Merge pull request #15 from jlac/main

Changes to calling and merging somatic variants
OpenOmics · Dec 5, 2023 · 460c4d1 · 460c4d1
2 parents 17fcce6 + c1d4092
commit 460c4d1
Show file tree

Hide file tree

Showing 11 changed files with 801 additions and 218 deletions.
diff --git a/VERSION b/VERSION
@@ -1 +1 @@
-0.5.0-alpha
+0.6.0
diff --git a/config/cluster.json b/config/cluster.json
@@ -170,7 +170,8 @@
     "glnexus": {
         "threads": "24",
         "mem": "240G",
-        "time": "4-00:00:00"
+        "time": "4-00:00:00",
+        "gres": "lscratch:256"
     },
     "gtype_refinement": {
         "threads": "2",
@@ -189,9 +190,9 @@
         "mem": "96G"
     },
     "hmftools_amber": {
-        "mem": "96G",
+        "mem": "365G",
         "time" : "48:00:00",
-        "threads": "24"
+        "threads": "56"
     },
     "hmftools_cobalt": {
         "mem": "96G",
@@ -203,6 +204,22 @@
         "time" : "48:00:00",
         "threads": "8"
     },
+    "somatic_purple_maf": {
+        "threads": "4",
+        "mem": "32G",
+        "time": "1-12:00:00",
+        "gres": "lscratch:32"
+    },
+    "purple_cohort_maf": {
+        "threads": "2",
+        "mem": "8G",
+        "time": "4:00:00"
+    },
+    "purple_cohort_maftools": {
+        "threads": "2",
+        "mem": "8G",
+        "time": "4:00:00"
+    },
     "inbreeding": {
         "threads": "8",
         "partition": "norm"
@@ -221,6 +238,11 @@
         "mem": "96G",
         "time": "1-00:00:00"
     },
+    "manta_somatic_filter": {
+        "threads": "4",
+        "mem": "72G",
+        "time": "1-00:00:00"
+    },
     "opencravat_germline_merge": {
         "threads": "4",
         "mem": "120G",
@@ -256,7 +278,7 @@
     "opencravat_germline": {
         "threads": "20",
         "mem": "48G",
-        "time": "12:00:00"
+        "time": "1-12:00:00"
     },
     "opencravat_somatic": {
         "threads": "32",
@@ -270,14 +292,19 @@
     },
     "octopus_somatic": {
         "threads": "28",
-        "mem": "48G",
-        "time": "1-18:00:00"
+        "mem": "120G",
+        "time": "2-12:00:00"
     },
     "octopus_merge": {
         "threads": "4",
         "mem": "48G",
         "time": "12:00:00"
     },
+    "octopus_filter": {
+        "threads": "4",
+        "mem": "48G",
+        "time": "12:00:00"
+    },
     "ped": {
         "threads": "2",
         "partition": "norm",
@@ -363,6 +390,11 @@
         "mem": "64G",
         "time": "24:00:00"
     },
+    "strelka_format": {
+        "threads": "16",
+        "mem": "64G",
+        "time": "24:00:00"
+    },
     "vcftools": {
         "threads": "2",
         "time": "12:00:00"

diff --git a/config/containers.json b/config/containers.json
@@ -4,7 +4,7 @@
         "genome-seek_trim_map": "docker://skchronicles/genome-seek_trim_map:v0.1.0",
         "genome-seek_somatic":  "docker://skchronicles/genome-seek_somatic:v0.1.0",
         "genome-seek_qc":       "docker://skchronicles/genome-seek_qc:v0.1.0",
-        "genome-seek_sv":       "docker://skchronicles/genome-seek_sv:v0.1.0",
+        "genome-seek_sv":       "docker://skchronicles/genome-seek_sv:v0.2.0",
         "genome-seek_cnv":      "docker://skchronicles/genome-seek_cnv:v0.2.0",
         "genome-seek_hla":      "docker://skchronicles/genome-seek_hla:v0.1.0",
         "base":                 "docker://skchronicles/ccbr_wes_base:v0.1.0",

diff --git a/config/genome.json b/config/genome.json
@@ -56,6 +56,9 @@
         ],
         "GENOME": "/data/OpenOmics/references/genome-seek/Homo_sapiens_assembly38.fasta",
         "GNOMAD": "/data/OpenOmics/references/genome-seek/GNOMAD/somatic-hg38-af-only-gnomad.hg38.vcf.gz",
+        "SLIVARGNOMAD": "/data/OpenOmics/references/genome-seek/slivar/gnomad.hg38.v2.zip",
+        "SLIVARGNOMAD3": "/data/OpenOmics/references/genome-seek/slivar/gnomad.hg38.genomes.v3.fix.zip",
+        "SLIVARTOPMED": "/data/OpenOmics/references/genome-seek/slivar/topmed.hg38.dbsnp.151.zip",
         "GISTIC_REFGENE": "/data/OpenOmics/references/genome-seek/gistic/hg38.UCSC.add_miR.160920.refgene.mat",
         "HLA_LA_GRAPH": "/data/OpenOmics/references/genome-seek/HLA-LA/graphs/PRG_MHC_GRCh38_withIMGT",
         "HMFTOOLS_REF_VERSION": "38",
@@ -88,6 +91,7 @@
             "recal5": ["chr11", "chr12", "chr17", "chr19", "chr22", "chrX"]
         },
         "MANTA_CALLREGIONS": "/data/OpenOmics/references/genome-seek/Manta/Homo_sapiens_assembly38_main_chromosomes.bed.gz",
+        "MANTA_FILTER_CHROMOSEQ_TRANSLOCATION": "/data/OpenOmics/references/genome-seek/chromoseq/chromoseq_translocations.bedpe",
         "OC_LIFTOVER": "hg38",
         "OC_MODULES": "/data/OpenOmics/references/genome-seek/OpenCRAVAT/modules",
         "OCTOPUS_SOMATIC_FOREST_MODEL": "/data/OpenOmics/references/genome-seek/Octopus/somatic.v0.7.4.forest",

diff --git a/config/modules.json b/config/modules.json
@@ -5,6 +5,7 @@
         "fastp": "fastp/0.23.1",
         "bowtie": "bowtie/2-2.4.5",
         "bcftools": "bcftools/1.13",
+        "bedtools": "bedtools/2.27.1",
         "canvas": "Canvas/1.40",
         "circos": "circos/0.69-9",
         "deepvariant": "deepvariant/1.4.0",
@@ -20,6 +21,7 @@
         "rlang": "R/4.2.0",
         "samblaster": "samblaster/0.1.25",
         "strelka": "strelka/2.9.10",
+        "svtools": "svtools/0.5.1",
         "python2": "python/2.7",
         "python3": "python/3.8",
         "fastq_screen": "fastq_screen/0.14.1",
@@ -28,6 +30,7 @@
         "gatk3": "GATK/3.8-1",
         "gatk4": "GATK/4.2.0.0",
         "multiqc": "multiqc/1.11",
+        "minimap2": "minimap2/2.26",
         "vcftools": "vcftools/0.1.16"
     }
 }
diff --git a/workflow/Snakefile b/workflow/Snakefile
@@ -309,6 +309,10 @@ rule all:
             join(workpath, "MANTA", "somatic", "{name}", "results", "variants", "somaticSV.vcf.gz"),
             name=provided(provided(tumors, call_somatic), call_sv),
         ),
+        expand( 
+            join(workpath, "MANTA", "somatic", "{name}", "results", "variants", "somaticSV.filtered.vcf"),
+            name=provided(provided(tumors, call_somatic), call_sv),
+        ),
         # Mutect2, call somatic variants,
         # scattered and merged on chroms
         # to speed up overall run time,
@@ -371,44 +375,19 @@ rule all:
             caller=provided(tn_somatic_callers, call_somatic),
             name=provided(tumorWnormal, call_somatic),
         ),
-        # Post-process somatic VCF, splits sites
-        # found in the tumor and normal sample, 
-        # and to merge callsets across callers,
+        # Post-process somatic VCF, 
+        # and merge callsets across callers,
         # only runs if provided --call-somatic 
         # @imported from rules/somatic.smk
         expand( 
-            join(workpath, "{caller}", "somatic", "{name}.{caller}.filtered.norm.tumor.vcf.gz"),
-            caller=provided(somatic_callers, call_somatic),
-            name=provided(tumors, call_somatic),
-        ),
-        # Conditionally add tumor-normal callers,
-        # MuSE and Strelka
-        expand( 
-            join(workpath, "{caller}", "somatic", "{name}.{caller}.filtered.norm.tumor.vcf.gz"),
-            caller=provided(tn_somatic_callers, call_somatic),
-            name=provided(tumorWnormal, call_somatic),
-        ),
-        expand( 
-            join(workpath, "{caller}", "somatic", "{name}.{caller}.filtered.norm.normal.vcf.gz"),
-            caller=provided(somatic_callers, call_somatic),
-            name=provided(tumorWnormal, call_somatic),
-        ),
-        # Conditionally add tumor-normal callers,
-        # MuSE and Strelka
-        expand( 
-            join(workpath, "{caller}", "somatic", "{name}.{caller}.filtered.norm.normal.vcf.gz"),
-            caller=provided(tn_somatic_callers, call_somatic),
-            name=provided(tumorWnormal, call_somatic),
-        ),
-        expand( 
-            join(workpath, "merged", "somatic", "{name}.merged.filtered.norm.tumor.vcf.gz"),
+            join(workpath, "merged", "somatic", "{name}.merged.filtered.norm.vcf.gz"),
             name=provided(tumors, call_somatic),
         ),
         # vcf2maf, create per-sample MAF file,
         # only runs if provided --call-somatic 
         # @imported from rules/somatic.smk
         expand( 
-            join(workpath, "merged", "somatic", "{name}.merged.filtered.norm.tumor.maf"),
+            join(workpath, "merged", "somatic", "{name}.merged.filtered.norm.maf"),
             name=provided(tumors, call_somatic),
         ),
         # vcf2maf, create cohort-level MAF file,
@@ -443,16 +422,29 @@ rule all:
         # @imported from rules/cnv.smk
         expand( 
             join(workpath, "hmftools", "amber", "{name}", "{name}.amber.baf.tsv"),
-            name=provided(provided(tumors, call_somatic), call_cnv),
+            name=provided(provided(tumors, call_somatic), call_cnv)
         ),
         expand( 
             join(workpath, "hmftools", "cobalt", "{name}", "{name}.cobalt.ratio.tsv"),
-            name=provided(provided(tumors, call_somatic), call_cnv),
+            name=provided(provided(tumors, call_somatic), call_cnv)
         ),
         expand( 
-            join(workpath, "hmftools", "purple", "{name}", "{caller}", "{name}.purple.cnv.somatic.tsv"),
-            name=provided(provided(tumors, call_somatic), call_cnv),
-            caller=provided(purple_callers, call_somatic),
+            join(workpath, "hmftools", "purple", "{name}", "{name}.purple.cnv.somatic.tsv"),
+            name=provided(provided(tumors, call_somatic), call_cnv)
+        ),
+        expand( 
+            join(workpath, "hmftools", "purple", "{name}", "{name}.purple.maf"),
+            name=provided(provided(tumors, call_somatic), call_cnv)
+        ),
+        # Maftools, create cohort-level summary plots from purple somatic VCF,
+        # only runs if provided --call-somatic AND --call-cnv
+        # @imported from rules/somatic.smk    
+        provided(
+            provided(
+                [join(workpath, "hmftools", "cohort_oncoplot.pdf")],
+                call_somatic
+            ),
+            call_cnv
         ),
         # OpenCRAVAT, somatic annotation, rank and score variants,
         # only runs if provided --open-cravat AND --call-somatic