Merge branch 'develop' into refactor/gatk_callers

Clinical-Genomics · Apr 28, 2022 · a6d98ef · a6d98ef
2 parents c0fcdd6 + 23773a5
commit a6d98ef
Show file tree

Hide file tree

Showing 44 changed files with 221 additions and 99 deletions.
diff --git a/BALSAMIC/config/analysis.json b/BALSAMIC/config/analysis.json
@@ -46,10 +46,14 @@
             "mutation": "somatic",
             "type": "SNV"
         },
-        "delly":{
+        "dellysv":{
             "mutation": "somatic",
             "type": "SV"
         },
+        "dellycnv":{
+            "mutation": "somatic",
+            "type": "CNV"
+        },
         "ascat": {
             "mutation": "somatic",
             "type": "SV"

diff --git a/BALSAMIC/constants/reference.py b/BALSAMIC/constants/reference.py
@@ -133,6 +133,30 @@
             "output_file": "delly_exclusion.tsv",
             "output_path": "genome",
         },
+        "delly_mappability": {
+            "url": "https://raw.githubusercontent.com/Clinical-Genomics/reference-files/86aab2d10c5ffc009bc8c68ad077ab7283d8fe06/cancer/references/GRCh38.delly.blacklist.gz",
+            "file_type": "text",
+            "gzip": False,
+            "genome_version": "hg38",
+            "output_file": "delly_mappability.gz",
+            "output_path": "genome",
+        },
+        "delly_mappability_gindex": {
+            "url": "https://raw.githubusercontent.com/Clinical-Genomics/reference-files/86aab2d10c5ffc009bc8c68ad077ab7283d8fe06/cancer/references/GRCh38.delly.blacklist.gz.gzi",
+            "file_type": "text",
+            "gzip": False,
+            "genome_version": "hg38",
+            "output_file": "delly_mappability.gz.gzi",
+            "output_path": "genome",
+        },
+        "delly_mappability_findex": {
+            "url": "https://raw.githubusercontent.com/Clinical-Genomics/reference-files/86aab2d10c5ffc009bc8c68ad077ab7283d8fe06/cancer/references/GRCh38.delly.blacklist.gz.fai",
+            "file_type": "text",
+            "gzip": False,
+            "genome_version": "hg38",
+            "output_file": "delly_mappability.gz.fai",
+            "output_path": "genome",
+        },
         "ascat_gccorrection": {
             "url": "https://raw.githubusercontent.com/Clinical-Genomics/reference-files/35465e2644f76f2d59427a9b379d34ecea71f259/cancer/references/hg38_SnpGcCorrections.tsv.gz",
             "file_type": "text",
@@ -287,6 +311,30 @@
             "output_file": "delly_exclusion.tsv",
             "output_path": "genome",
         },
+        "delly_mappability": {
+            "url": "https://raw.githubusercontent.com/Clinical-Genomics/reference-files/86aab2d10c5ffc009bc8c68ad077ab7283d8fe06/cancer/references/GRCh37.delly.blacklist.gz",
+            "file_type": "text",
+            "gzip": False,
+            "genome_version": "hg19",
+            "output_file": "delly_mappability.gz",
+            "output_path": "genome",
+        },
+        "delly_mappability_gindex": {
+            "url": "https://raw.githubusercontent.com/Clinical-Genomics/reference-files/86aab2d10c5ffc009bc8c68ad077ab7283d8fe06/cancer/references/GRCh37.delly.blacklist.gz.gzi",
+            "file_type": "text",
+            "gzip": False,
+            "genome_version": "hg19",
+            "output_file": "delly_mappability.gz.gzi",
+            "output_path": "genome",
+        },
+        "delly_mappability_findex": {
+            "url": "https://raw.githubusercontent.com/Clinical-Genomics/reference-files/86aab2d10c5ffc009bc8c68ad077ab7283d8fe06/cancer/references/GRCh37.delly.blacklist.gz.fai",
+            "file_type": "text",
+            "gzip": False,
+            "genome_version": "hg19",
+            "output_file": "delly_mappability.gz.fai",
+            "output_path": "genome",
+        },
         "ascat_gccorrection": {
             "url": "https://raw.githubusercontent.com/Clinical-Genomics/reference-files/12a6c760fd542c02de2cda286b6245e46f4b6a97/cancer/references/GRCh37_SnpGcCorrections.tsv.gz",
             "file_type": "text",

diff --git a/BALSAMIC/constants/workflow_params.py b/BALSAMIC/constants/workflow_params.py
@@ -64,13 +64,27 @@
         "sequencing_type": ["targeted", "wgs"],
         "workflow_solution": ["BALSAMIC"],
     },
-    "delly": {
+    "haplotypecaller": {
+        "mutation": "germline",
+        "type": "SNV",
+        "analysis_type": ["paired", "single"],
+        "sequencing_type": ["targeted"],
+        "workflow_solution": ["BALSAMIC"],
+    },
+    "dellysv": {
         "mutation": "somatic",
         "type": "SV",
         "analysis_type": ["paired", "single"],
         "sequencing_type": ["targeted", "wgs"],
         "workflow_solution": ["BALSAMIC"],
     },
+    "dellycnv": {
+        "mutation": "somatic",
+        "type": "CNV",
+        "analysis_type": ["single"],
+        "sequencing_type": ["targeted", "wgs"],
+        "workflow_solution": ["BALSAMIC"],
+    },
     "ascat": {
         "mutation": "somatic",
         "type": "CNV",

diff --git a/BALSAMIC/snakemake_rules/annotation/rankscore.rule b/BALSAMIC/snakemake_rules/annotation/rankscore.rule
@@ -19,11 +19,11 @@ rule genmod_score_vardict:
   threads:
     get_threads(cluster_config, 'genmod_score_vardict')
   message:
-    ("Score annotated vardict variants using genmod"
-    "and compress vcf using bcftools on {params.case_name}")
+    ("Scoring annotated vardict variants using genmod for {params.case_name}")
   shell:
     """
 genmod score -r -c {input.rankscore} {input.vcf} | \
+
 bcftools view -o {output.vcf_pass} -O z;
 
 tabix -p vcf -f {output.vcf_pass};

diff --git a/BALSAMIC/snakemake_rules/annotation/varcaller_filter_tumor_normal.rule b/BALSAMIC/snakemake_rules/annotation/varcaller_filter_tumor_normal.rule
@@ -28,7 +28,7 @@ rule bcftools_filter_vardict_tumor_normal:
   threads:
     get_threads(cluster_config, 'bcftools_filter_vardict_tumor_normal')
   message:
-    "Filtering vardict tumor-normal annotated variants using bcftools on {params.case_name}"
+    "Filtering vardict tumor-normal annotated variants using bcftools for {params.case_name}"
   shell:
     """
 bcftools view {input.vcf} \
@@ -68,7 +68,7 @@ rule bcftools_filter_tnhaplotyper_tumor_normal:
   threads:
     get_threads(cluster_config, 'bcftools_filter_tnhaplotyper_tumor_normal')
   message:
-    "Filtering tnhaplotyper tumor-normal annotated variants using bcftools on {params.case_name}"
+    "Filtering tnhaplotyper tumor-normal annotated variants using bcftools for {params.case_name}"
   shell:
     """
 bcftools view {input.vcf} \
@@ -102,7 +102,7 @@ rule bcftools_filter_TNscope_umi_tumor_normal:
   threads:
     get_threads(cluster_config, 'bcftools_filter_TNscope_umi_tumor_normal')
   message:
-    "Filtering TNscope_umi tumor-normal annotated variants using bcftools on {params.case_name}"
+    "Filtering TNscope_umi tumor-normal annotated variants using bcftools for {params.case_name}"
   shell:
     """
 bcftools view {input.vcf} \

diff --git a/BALSAMIC/snakemake_rules/annotation/varcaller_filter_tumor_only.rule b/BALSAMIC/snakemake_rules/annotation/varcaller_filter_tumor_only.rule
@@ -66,7 +66,7 @@ rule bcftools_filter_tnhaplotyper_tumor_only:
   threads:
     get_threads(cluster_config, 'bcftools_filter_tnhaplotyper_tumor_only')
   message:
-    "Filtering tnhaplotyper tumor-only annotated variants using bcftools on {params.case_name}"
+    "Filtering tnhaplotyper tumor-only annotated variants using bcftools for {params.case_name}"
   shell:
     """
 bcftools view {input.vcf} \
@@ -100,7 +100,7 @@ rule bcftools_filter_TNscope_umi_tumor_only:
   threads:
     get_threads(cluster_config, 'bcftools_filter_TNscope_umi_tumor_only')
   message:
-    "Filtering TNscope_umi tumor-only annotated variants using bcftools on {params.case_name}"
+    "Filtering TNscope_umi tumor-only annotated variants using bcftools for {params.case_name}"
   shell:
     """
 bcftools view {input.vcf} \

diff --git a/BALSAMIC/snakemake_rules/annotation/varcaller_sv_filter.rule b/BALSAMIC/snakemake_rules/annotation/varcaller_sv_filter.rule
@@ -18,7 +18,7 @@ rule bcftools_filter_svdb:
   threads:
     get_threads(cluster_config, "bcftools_filter_svdb")
   message:
-    "Filtering svdb merged Manta and Delly results for PASS variants using bcftools for sample '{params.case_name}' "
+    "Filtering merged structural and copy number variants using bcftools for {params.case_name}"
   shell:
     """
 bcftools view --threads {threads} -f .,PASS -o {output.vcf_pass_svdb} -O z {input.vcf};

diff --git a/BALSAMIC/snakemake_rules/annotation/varcaller_wgs_filter_tumor_normal.rule b/BALSAMIC/snakemake_rules/annotation/varcaller_wgs_filter_tumor_normal.rule
@@ -26,7 +26,7 @@ rule bcftools_filter_tnscope_tumor_normal:
   threads:
     get_threads(cluster_config, 'bcftools_filter_tnscope_tumor_normal')
   message:
-    "Filtering wgs tumor-normal tnscope annotated variants using bcftools on {params.case_name}"
+    "Filtering WGS tumor-normal tnscope annotated variants using bcftools for {params.case_name}"
   shell:
     """
 bcftools view {input.vcf} \
@@ -65,7 +65,7 @@ rule bcftools_filter_tnhaplotyper_tumor_normal:
   threads: 
     get_threads(cluster_config, 'bcftools_filter_tnhaplotyper_tumor_normal')
   message:
-    "Filtering wgs tumor-normal tnhaplotyper annotated variants using bcftools on {params.case_name}"
+    "Filtering WGS tumor-normal tnhaplotyper annotated variants using bcftools for {params.case_name}"
   shell:
     """
 grep -v '^@' {input.wgs_calling_file} > {input.wgs_calling_file}.bed

diff --git a/BALSAMIC/snakemake_rules/annotation/varcaller_wgs_filter_tumor_only.rule b/BALSAMIC/snakemake_rules/annotation/varcaller_wgs_filter_tumor_only.rule
@@ -28,7 +28,7 @@ rule bcftools_filter_tnscope_tumor_only:
   threads: 
     get_threads(cluster_config, 'bcftools_filter_tnscope_tumor_only')
   message:
-    "Filtering wgs tumor-only tnscope annotated variants using bcftools on {params.case_name}"
+    "Filtering WGS tumor-only tnscope annotated variants using bcftools for {params.case_name}"
   shell:
     """
 grep -v '^@' {input.wgs_calling_file} > {input.wgs_calling_file}.bed
@@ -72,7 +72,7 @@ rule bcftools_filter_tnhaplotyper_tumor_only:
   threads: 
     get_threads(cluster_config, 'bcftools_filter_tnhaplotyper_tumor_only')
   message:
-    "Filtering wgs tumor-only tnhaplotyper annotated variants using bcftools on {params.case_name}"
+    "Filtering WGS tumor-only tnhaplotyper annotated variants using bcftools for {params.case_name}"
   shell:
     """
 grep -v '^@' {input.wgs_calling_file} > {input.wgs_calling_file}.bed

diff --git a/BALSAMIC/snakemake_rules/annotation/vcf2cytosure_convert.rule b/BALSAMIC/snakemake_rules/annotation/vcf2cytosure_convert.rule
@@ -17,7 +17,7 @@ rule vcf2cytosure_convert:
     params:
         case_name = config["analysis"]["case_id"],
         housekeeper_id = {"id": config["analysis"]["case_id"], "tags": "cnv-somatic"},
-    message: "Convert VCF file with CNVs to the .CGH format using vcf2cytosure for sample {params.case_name}"
+    message: "Converting CNVs from VCF to the CGH format using vcf2cytosure for {params.case_name}"
     shell:
         """
 vcf2cytosure --vcf {input.cnv_vcf} --cn {input.cnv_cnr} --out {output.cgh_file} --bins 1

diff --git a/BALSAMIC/snakemake_rules/annotation/vep.rule b/BALSAMIC/snakemake_rules/annotation/vep.rule
@@ -25,7 +25,7 @@ rule vep_somatic_snv:
   threads:
     get_threads(cluster_config, "vep_somatic_snv")
   message:
-    "Running vep annotation on {params.message_text}"
+    "Running vep annotation for single nuceotide variants on {params.message_text}"
   shell:
     """
 vep_path=$(dirname $(readlink -f $(which vep)));
@@ -69,7 +69,7 @@ rule vep_somatic_sv:
   threads:
     get_threads(cluster_config, "vep_somatic_sv")
   message:
-    "Running vep annotation on {params.message_text}"
+    "Running vep annotation for structural and copy number variants on {params.message_text}"
   shell:
     """
 vep_path=$(dirname $(readlink -f $(which vep)));
@@ -106,7 +106,7 @@ rule tmb_calculation:
   threads:
     get_threads(cluster_config, "vep")
   message:
-    "Calculating TMB for {params.message_text}" 
+    "Calculating TMB score for {params.message_text}"
   shell:
     """
 mkdir -p {params.tmpdir};
@@ -163,7 +163,7 @@ rule vep_germline:
   threads:
     get_threads(cluster_config, 'vep_germline')
   message:
-    "Running vep annotation on {params.sample}"
+    "Running vep annotation on germline variants for {params.sample}"
   shell:
       """
 vep_path=$(dirname $(readlink -f $(which vep)));

diff --git a/BALSAMIC/snakemake_rules/umi/generate_AF_tables.rule b/BALSAMIC/snakemake_rules/umi/generate_AF_tables.rule
@@ -19,7 +19,7 @@ rule bcftools_query_generatebackgroundaf_umitable:
     threads: 
         get_threads(cluster_config, "bcftools_query_generatebackgroundaf_umitable")
     message: 
-        "Creating Allelic frequency table from VCF file for sample {params.case_name}"
+        "Creating Allelic frequency table from VCF file for {params.case_name}"
     shell:
         """
 bcftools query \

diff --git a/BALSAMIC/snakemake_rules/umi/mergetype_normal_umi.rule b/BALSAMIC/snakemake_rules/umi/mergetype_normal_umi.rule
@@ -21,12 +21,14 @@ rule mergeBam_normal_umiconsensus:
   threads:
     get_threads(cluster_config, "mergeBam_normal_umiconsensus")
   message:
-    ("Replace ReadGroups using picard for normal sample {params.sample} "
-    "and convert bam to cram format")
+    ("Replacing ReadGroups using picard and converting from bam to cram format for {params.sample}")
   shell:
         """
 picard AddOrReplaceReadGroups {params.picard} INPUT={input.bam} OUTPUT={output.bam}; 
+
 samtools index {output.bam};
+
 samtools view -h -T {input.fasta} --threads {threads} -C -o {output.cram} {output.bam}; 
+
 samtools index {output.cram}; 
         """
diff --git a/BALSAMIC/snakemake_rules/umi/mergetype_tumor_umi.rule b/BALSAMIC/snakemake_rules/umi/mergetype_tumor_umi.rule
@@ -21,8 +21,7 @@ rule mergeBam_tumor_umiconsensus:
     threads: 
         get_threads(cluster_config, "mergeBam_tumor_umiconsensus")
     message:
-        ("Replace ReadGroups using picard for tumor sample {params.sample} "
-        "and convert bam to cram")
+        ("Replacing ReadGroups using picard and converting from bam to cram for {params.sample}")
     shell:
         """
 picard AddOrReplaceReadGroups {params.picard} INPUT={input.bam} OUTPUT={output.bam};

diff --git a/BALSAMIC/snakemake_rules/umi/qc_umi.rule b/BALSAMIC/snakemake_rules/umi/qc_umi.rule
@@ -18,7 +18,7 @@ rule picard_umiaware:
     threads:
         get_threads(cluster_config, "picard_umiaware")
     message:
-        "Picard Umiaware mark dups for sample {params.sample_id}"
+        "Marking duplicates using Picardtools with UmiAware for {params.sample_id}"
     shell:
         """
 picard UmiAwareMarkDuplicatesWithMateCigar \
@@ -47,7 +47,7 @@ rule picard_collecthsmetrics_umi:
     threads: 
         get_threads(cluster_config, "CollectHsMetrics")
     message:
-        "Collect HSmetrics using Picardtools for {params.sample_id}"
+        "Collecting HSmetrics using Picardtools for {params.sample_id}"
     shell:
         """
 picard BedToIntervalList \

diff --git a/BALSAMIC/snakemake_rules/umi/sentieon_consensuscall.rule b/BALSAMIC/snakemake_rules/umi/sentieon_consensuscall.rule
@@ -22,7 +22,7 @@ rule sentieon_consensuscall_umi:
     threads: 
         get_threads(cluster_config, "sentieon_consensuscall_umi")
     message: 
-        "Consensus molecule creation using sentieon for sample {params.sample_id}"
+        "Calling consensus molecules using sentieon for {params.sample_id}"
     shell:
         """
 export LD_PRELOAD={params.sentieon_install_dir}/lib/libjemalloc.so.1
@@ -63,7 +63,7 @@ rule sentieon_bwa_umiconsensus:
     threads: 
         get_threads(cluster_config, "sentieon_bwa_umiconsensus")
     message:
-         "Mapping of consensus reads with the sentieon bwa mem, sorting for sample {params.sample_id}"
+         "Mapping consensus reads and sorting using sentieon bwa-mem for {params.sample_id}"
     shell:
         """
 export LD_PRELOAD={params.sentieon_install_dir}/lib/libjemalloc.so.1
@@ -106,7 +106,7 @@ rule sentieon_consensusfilter_umi:
     threads:
         get_threads(cluster_config, "sentieon_consensusfilter_umi")
     message:
-        "Filtering consensus reads based on XZ tag for sample {params.sample_id}"
+        "Filtering consensus reads based on XZ tag for {params.sample_id}"
     shell:
         """
 samtools view -h {input} | \

diff --git a/BALSAMIC/snakemake_rules/umi/sentieon_umiextract.rule b/BALSAMIC/snakemake_rules/umi/sentieon_umiextract.rule
@@ -22,7 +22,7 @@ rule sentieon_umiextract:
     threads:
         get_threads(cluster_config, "sentieon_umiextract")
     message: 
-        "UMI tag extraction using sentieon for sample {params.sample_id}"
+        "Extracing UMI tags using sentieon for {params.sample_id}"
     shell:
         """
 export LD_PRELOAD={params.sentieon_install_dir}/lib/libjemalloc.so.1
@@ -59,7 +59,7 @@ rule sentieon_bwa_umiextract:
     threads: 
         get_threads(cluster_config, "sentieon_bwa_umiextract")
     message: 
-        "Aligning of UMI extracted reads with sentieon bwa mem, sorting for sample {params.sample_id}"
+        "Aligning UMI extracted reads and sorting using sentieon bwa-mem for {params.sample_id}"
     shell:
         """
 export LD_PRELOAD={params.sentieon_install_dir}/lib/libjemalloc.so.1

diff --git a/BALSAMIC/snakemake_rules/umi/sentieon_varcall_tnscope.rule b/BALSAMIC/snakemake_rules/umi/sentieon_varcall_tnscope.rule
@@ -33,7 +33,7 @@ rule sentieon_tnscope_umi:
     threads: 
         get_threads(cluster_config, "sentieon_tnscope_umi")
     message: 
-        "Calling SNVs using TNscope for sample {params.tumor}"
+        "Calling single nucleotide variants using TNscope for {params.tumor}"
     shell:
         """
 mkdir -p {params.tmpdir};

diff --git a/BALSAMIC/snakemake_rules/umi/sentieon_varcall_tnscope_tn.rule b/BALSAMIC/snakemake_rules/umi/sentieon_varcall_tnscope_tn.rule
@@ -30,12 +30,12 @@ rule sentieon_tnscope_umi_tn:
         pcr_model = params.common.pcr_model,
         padding = params.tnscope_umi.padding,
         tumor = "TUMOR",
-        normal = "NORMAL"
+        normal = "NORMAL",
+        case_name= config["analysis"]["case_id"]
     threads: 
         get_threads(cluster_config, "sentieon_tnscope_umi")
     message: 
-        "Calling SNVs using TNscope for sample: {params.tumor}"
-        " versus sample {params.normal}"
+        "Calling single nucleotide variants using TNscope for {params.case_name}"
     shell:
         """
 mkdir -p {params.tmpdir};

diff --git a/BALSAMIC/snakemake_rules/variant_calling/cnvkit_paired.rule b/BALSAMIC/snakemake_rules/variant_calling/cnvkit_paired.rule
@@ -35,7 +35,7 @@ rule cnvkit_paired:
         sample_id = "TUMOR",
         genome = GENOME_VERSION
     message:
-        "Run CNVkit pipeline for sample {params.case_name} while tumor purity/ploidy calculated using PureCN"
+        "Calling CNVs using CNVkit and calculating tumor purity/ploidy using PureCN for {params.case_name}"
     shell:
         """
 mkdir -p {params.tmpdir};

diff --git a/BALSAMIC/snakemake_rules/variant_calling/cnvkit_single.rule b/BALSAMIC/snakemake_rules/variant_calling/cnvkit_single.rule
@@ -44,8 +44,7 @@ rule cnvkit_single:
         genome_version = GENOME_VERSION,
         pon = " " if get_pon_cnn(config) is None else get_pon_cnn(config)
     message:
-        ("Run CNVkit pipeline for sample {params.case_name},"
-        "while tumor purity/ploidy calculated using PureCN")
+        ("Calling CNVs using CNVkit and calculating tumor purity/ploidy using PureCN for {params.case_name}")
     shell:
       """
 mkdir -p {params.tmpdir};