BINF-494: add individual pon workflow (#19)

* BINF-494: add individual pon workflow
* BINF-494: fix workflow
* BINF-494: update base image
* BINF-494: add option to skip download some reference for pon
* BINF-494: fix prefix
* BINF-494: fix download when skipping
* BINF-494: fix workflow input source

example/gatk4.2.4.1_mutect2_tumor_only.json

@@ -1,18 +1,14 @@
 {
     "has_normal": [],
-    "output_prefix": "0708_test_full",
+    "output_prefix": "pon_test_full",
     "java_heap": "30G",
     "chunk_java_heap": "8G",
     "nthreads": 4,
-    "reference": {"class": "File", "path": "/mnt/SCRATCH/reference/GRCh38.d1.vd1.fa"},
-    "reference_image": {"class": "File", "path": "/mnt/SCRATCH/reference/GRCh38.d1.vd1.fa.img"},
-    "common_variant_reference": {"class": "File", "path": "/mnt/SCRATCH/reference/gnomad.genomes.r2.1.1.sites.liftover_grch38.pass.no_decoy.af_only.biallelicvcf.gz"},
+    "reference": {"class": "File", "path": "/mnt/SCRATCH/reference/hwf/GRCh38.d1.vd1.fa"},
     "intervals": {"class": "File", "path": "/mnt/SCRATCH/reference/intervals.bed"},
-    "tumor_bam": {"class": "File", "path": "/mnt/SCRATCH/playground/C500.TCGA-HQ-A5ND-01A-11D-A26M-08.4_gdc_realn.bam"},
-    "f1r2_tar_gz": true,
-    "bam_output": true,
-    "genotype_pon_sites": true,
-    "germline_resource": {"class": "File", "path": "/mnt/SCRATCH/reference/gnomad.genomes.r2.1.1.sites.liftover_grch38.pass.no_decoy.af_only.vcf.gz"},
-    "panel_of_normals": {"class": "File", "path": "/mnt/SCRATCH/reference/gatk4.1.2.tcga.4136.pon.vcf.gz"},
-    "call_on_all": true
+    "input": [{"class": "File", "path": "/mnt/SCRATCH/playground/bam_files/C500.TCGA-HQ-A5ND-01A-11D-A26M-08.4_gdc_realn.bam"}],
+    "f1r2_tar_gz": false,
+    "bam_output": false,
+    "call_on_all": false,
+    "max_mnp_distance": 0
 }

gatk4-mutect2-calling-cwl/Dockerfile

@@ -1,4 +1,4 @@
-FROM quay.io/ncigdc/bio-alpine:py36
+FROM quay.io/ncigdc/bio-alpine:py36-8dee591
 
 ARG WORKFLOW
 

gatk4-mutect2-calling-cwl/gpas_gatk4.2.4.1_mutect2_workflow.cwl

@@ -14,6 +14,7 @@ requirements:
 inputs:
 ###CONDITIONAL_INPUTS###
   has_normal: int[]
+  pon_calling: int[]
 ###BIOCLIENT_INPUTS###
   bioclient_config: File
   tumor_gdc_id: string
@@ -207,6 +208,7 @@ steps:
   preparation:
     run: ../utils-cwl/subworkflow/preparation_workflow.cwl
     in:
+      pon_calling: pon_calling
       bioclient_config: bioclient_config
       has_normal: has_normal
       tumor_gdc_id: tumor_gdc_id

gatk4-mutect2-tumor-only-cwl/Dockerfile

@@ -1,4 +1,4 @@
-FROM quay.io/ncigdc/bio-alpine:py36
+FROM quay.io/ncigdc/bio-alpine:py36-8dee591
 
 ARG WORKFLOW
 

gatk4-mutect2-tumor-only-cwl/gpas_gatk4.2.4.1_mutect2_tumor_only_workflow.cwl

@@ -14,6 +14,7 @@ requirements:
 inputs:
 ###CONDITIONAL_INPUTS###
   has_normal: int[]
+  pon_calling: int[]
 ###BIOCLIENT_INPUTS###
   bioclient_config: File
   tumor_gdc_id: string
@@ -207,6 +208,7 @@ steps:
   preparation:
     run: ../utils-cwl/subworkflow/preparation_workflow.cwl
     in:
+      pon_calling: pon_calling
       bioclient_config: bioclient_config
       has_normal: has_normal
       tumor_gdc_id: tumor_gdc_id

gatk4-pon-cwl/Dockerfile

@@ -0,0 +1,17 @@
+FROM quay.io/ncigdc/bio-alpine:py36-8dee591
+
+ARG WORKFLOW
+
+COPY ./${WORKFLOW} /opt
+# TODO: Copy additional directories as needed
+COPY ./tools /tools
+COPY ./utils-cwl /utils-cwl
+COPY ./subworkflows /subworkflows
+
+RUN make -C /opt init-pip
+
+WORKDIR /opt
+
+ENTRYPOINT ["make"]
+
+CMD ["pack"]

gatk4-pon-cwl/Makefile

@@ -0,0 +1,72 @@
+# quay repo not git repo
+REPO = gatk4-pon-cwl
+# TODO: Replace me
+ENTRY_CWL = "gpas_gatk4.2.4.1_mutect2_pon_workflow.cwl"
+
+VERSION := $(shell date -u +"%Y%m%dT%H%MZ")
+COMMIT_HASH:=$(shell git rev-parse HEAD)
+
+DOCKER_REPO := quay.io/ncigdc
+DOCKER_IMAGE_COMMIT := ${DOCKER_REPO}/${REPO}:${COMMIT_HASH}
+DOCKER_IMAGE := ${DOCKER_REPO}/${REPO}:${VERSION}
+
+.PHONY: version version-* name
+name:
+	@echo ${WORKFLOW_NAME}
+
+version:
+	@echo --- VERSION: ${VERSION} ---
+
+version-docker:
+	@echo ${DOCKER_IMAGE_COMMIT}
+	@echo ${DOCKER_IMAGE}
+
+.PHONY: docker-*
+docker-login:
+	@echo
+	docker login -u="${QUAY_USERNAME}" -p="${QUAY_PASSWORD}" quay.io
+
+
+.PHONY: build build-* clean init init-* lint requirements run version
+init: init-pip
+
+init-pip:
+	@echo
+	@echo -- Installing pip packages --
+	pip3 install --no-cache-dir cwltool==1.0.20180306163216
+
+.PHONY: pack pack-%
+pack:
+	@python -m cwltool --pack "${ENTRY_CWL}"
+
+run:
+	@docker run --rm ${DOCKER_IMAGE_COMMIT} pack ENTRY_CWL=/opt/${ENTRY_CWL}
+
+.PHONY: validate validate-*
+validate:
+	@python -m cwltool --validate "${ENTRY_CWL}"
+
+validate-docker:
+	@docker run --rm ${DOCKER_IMAGE_COMMIT} validate ENTRY_CWL=/opt/${ENTRY_CWL}
+
+.PHONY: build build-*
+build: build-docker
+
+build-docker:
+	@echo
+	@echo -- Building docker --
+	docker build .. \
+		--file ./Dockerfile \
+		--build-arg WORKFLOW=${WORKFLOW_NAME} \
+		-t "${DOCKER_IMAGE_COMMIT}"
+
+
+.PHONY: publish publish-release
+publish:
+	docker push ${DOCKER_IMAGE_COMMIT}
+
+publish-staging: publish
+
+publish-release: publish
+	docker tag ${DOCKER_IMAGE_COMMIT} ${DOCKER_IMAGE}
+	docker push ${DOCKER_IMAGE}