Add facets-suite somatic variant caller

config/day_profiles/local/templates/rule_config.yaml

@@ -196,6 +196,13 @@ deepsomatic:
     numa: " OMP_NUM_THREADS=8 OMP_PROC_BIND=close OMP_PLACES=threads OMP_PROC_BIND=TRUE OMP_DYNAMIC=TRUE OMP_MAX_ACTIVE_LEVELS=1 OMP_SCHEDULE=dynamic OMP_WAIT_POLICY=ACTIVE "
     dvsom_conda: "../envs/vanilla_v0.1.yaml"
 
+facets:
+    threads: 2
+    env_yaml: "../envs/vanilla_v0.1.yaml"
+    container: "docker://mskcc/facets-suite:2.2.1"
+    partition: "i8"
+    mem_mb: 8000
+
 duphold:
     threads: 7
     env_yaml: "../envs/duphold_v0.1.yaml"

config/day_profiles/slurm/templates/rule_config.yaml

@@ -195,6 +195,13 @@ deepsomatic:
     numa: " OMP_THREADS=64 OMP_PROC_BIND=close OMP_PLACES=threads  OMP_DYNAMIC=true OMP_MAX_ACTIVE_LEVELS=1 OMP_SCHEDULE=dynamic OMP_WAIT_POLICY=ACTIVE "
     dvsom_conda: "../envs/vanilla_v0.1.yaml"
 
+facets:
+    threads: 8
+    env_yaml: "../envs/vanilla_v0.1.yaml"
+    container: "docker://mskcc/facets-suite:2.2.1"
+    partition: "i192,i128,i192mem"
+    mem_mb: 16000
+
 
 
 duphold:

config/global.yaml

@@ -4,6 +4,7 @@ profile_name: "global"  # this is over-ridden when profile specific config is in
 
 snv_callers_somatic:
   - "senttn"
+  - "facets"
 
 
 logs:

workflow/Snakefile

@@ -265,6 +265,7 @@ include: "rules/deepvariant_1_5.smk"
 include: "rules/deepvariant_1_9.smk"
 include: "rules/deepvariant_ug.smk"
 include: "rules/deepsomatic.smk"
+include: "rules/facets_suite.smk"
 include: "rules/doppel_mrkdups.smk"
 include: "rules/duphold.smk"
 include: "rules/dysgu_sv.smk"

workflow/rules/facets_suite.smk

@@ -0,0 +1,74 @@
+import os
+
+##### facets-suite somatic variant caller
+# ---------------------------
+
+rule facets_suite:
+    wildcard_constraints:
+        sample=TUMORS_REGEX
+    input:
+        tumor_cram=get_somcall_tumor_cram,
+        tumor_crai=get_somcall_tumor_crai,
+        normal_cram=get_somcall_normal_cram,
+        normal_crai=get_somcall_normal_crai,
+        ref_fa=lambda wc: config["supporting_files"]["files"]["huref"]["fasta"]["name"],
+        ref_fai=lambda wc: config["supporting_files"]["files"]["huref"]["fasta"]["name"] + ".fai",
+    output:
+        cncf=MDIR + "{sample}/align/{alnr}/snv/facets/{sample}.{alnr}.facets.cncf.txt",
+        seg=MDIR + "{sample}/align/{alnr}/snv/facets/{sample}.{alnr}.facets.seg",
+        vcfgz=MDIR + "{sample}/align/{alnr}/snv/facets/{sample}.{alnr}.facets.vcf.gz",
+        vcftbi=MDIR + "{sample}/align/{alnr}/snv/facets/{sample}.{alnr}.facets.vcf.gz.tbi",
+    log:
+        MDIR + "{sample}/align/{alnr}/snv/facets/log/{sample}.{alnr}.facets.log",
+    threads: config['facets']['threads']
+    container:
+        config['facets']['container']
+    resources:
+        vcpu=config['facets']['threads'],
+        threads=config['facets']['threads'],
+        partition=config['facets']['partition'],
+        mem_mb=config['facets']['mem_mb'],
+    params:
+        cluster_sample=ret_sample,
+    shell:
+        r"""
+        set -euo pipefail
+        mkdir -p "$(dirname {output.cncf})"
+
+        run_facets_wrapper.R \
+            --tumor {input.tumor_cram} \
+            --normal {input.normal_cram} \
+            --reference {input.ref_fa} \
+            --outdir $(dirname {output.cncf}) \
+            --name {wildcards.sample}.{wildcards.alnr} \
+            --vcf {output.vcfgz} \
+            --cncf {output.cncf} \
+            --seg {output.seg} >> {log} 2>&1
+
+        tabix -f -p vcf {output.vcfgz} >> {log} 2>&1
+        """
+
+
+rule produce_facets_suite_vcf:  # Target rule to gather facets outputs
+    wildcard_constraints:
+        sample=TUMORS_REGEX
+    input:
+        vcftb=expand(
+            MDIR + "{sample}/align/{alnr}/snv/facets/{sample}.{alnr}.facets.vcf.gz",
+            sample=TN_TUMOR_SAMPS,
+            alnr=ALIGNERS,
+        ),
+        vcftbi=expand(
+            MDIR + "{sample}/align/{alnr}/snv/facets/{sample}.{alnr}.facets.vcf.gz.tbi",
+            sample=TN_TUMOR_SAMPS,
+            alnr=ALIGNERS,
+        ),
+    output:
+        "gatheredall.facets",
+    threads: 1
+    log:
+        "gatheredall.facets.log",
+    params:
+        cluster_sample=ret_sample,
+    shell:
+        "touch {output}"