Skip to content

Instantly share code, notes, and snippets.

@thanhpp
Last active September 30, 2020 02:17
Show Gist options
  • Save thanhpp/831b962d19ffa5ee450df34acbe66c13 to your computer and use it in GitHub Desktop.
Save thanhpp/831b962d19ffa5ee450df34acbe66c13 to your computer and use it in GitHub Desktop.
{{host}}:{{port}}/dag
{
"Error": {
"Code": 200,
"Message": "Created DAG"
},
"Data": {
"dag_id": 1,
"dag_name": "#main",
"user_id": 1,
"tasks": [
{
"task_id": "1-0",
"dag_id": 1,
"task_name": "#AddReplaceReadGroups.cwl",
"task_workflow_name": "AddReplaceReadGroups",
"docker_images": [
"vinbdi/biotools"
],
"requirement": "",
"base_command": "picard AddOrReplaceReadGroups ",
"tool_id": 0,
"parameters": [
""
],
"values": [
{
"name": "common_basename",
"workflow_name": "output_basename",
"from": "",
"type": "string",
"prefix": " CREATE_INDEX=true RGID=RG_mother RGLB=lib_mother RGPL=illumina RGPU=unit_mother RGSM=",
"null_flag": false,
"have_regex": false,
"default_value": [
"mother"
]
},
{
"name": "input_bam",
"workflow_name": "output",
"from": "SortBam",
"type": "File",
"prefix": "I=",
"null_flag": false,
"have_regex": true,
"default_value": [
"*.bam"
]
},
{
"name": "output_add_replace_rdg",
"workflow_name": "output_basename",
"from": "",
"type": "string",
"prefix": " O=",
"null_flag": false,
"have_regex": false,
"default_value": [
"mother.grp.sorted.md.bam"
]
}
],
"outputs": [
{
"name": "output",
"datatype": "single",
"type": "File",
"partern": [
"*.grp.sorted.md.bam"
],
"regex": [
"*.grp.sorted.md.bam"
],
"source": ""
}
],
"parrent_task_id": [
"1-7"
],
"children_task_id": [
"1-6"
]
},
{
"task_id": "1-1",
"dag_id": 1,
"task_name": "#AlignBwaMem.cwl",
"task_workflow_name": "AlignmentReads",
"docker_images": [
"vinbdi/biotools"
],
"requirement": "",
"base_command": "bwa mem ",
"tool_id": 0,
"parameters": [
""
],
"values": [
{
"name": "indexed_reference_fasta",
"workflow_name": "indexed_reference_fasta",
"from": "",
"type": "File",
"prefix": "-M -Y -t 16 -K 100000000 -p ",
"null_flag": false,
"have_regex": false,
"default_value": [
"/home/ubuntu/references/Homo_sapiens_assembly38.fasta"
]
},
{
"name": "input_fastqs",
"workflow_name": "input_fastqs",
"from": "",
"type": "File",
"prefix": " ",
"null_flag": false,
"have_regex": false,
"default_value": [
"/home/ubuntu/input/mother_R1.fq.gz",
"/home/ubuntu/input/mother_R2.fq.gz"
]
},
{
"name": "output_aligned_reads",
"workflow_name": "output_basename",
"from": "",
"type": "File",
"prefix": " -o ",
"null_flag": false,
"have_regex": false,
"default_value": [
"mother.sam"
]
}
],
"outputs": [
{
"name": "output",
"datatype": "single",
"type": "File",
"partern": [
"*.sam"
],
"regex": [
"*.sam"
],
"source": ""
}
],
"parrent_task_id": null,
"children_task_id": [
"1-3"
]
},
{
"task_id": "1-2",
"dag_id": 1,
"task_name": "#ApplyBQSR.cwl",
"task_workflow_name": "ApplyBQSR",
"docker_images": [
"vinbdi/biotools"
],
"requirement": "",
"base_command": "gatk ApplyBQSR ",
"tool_id": 0,
"parameters": [
" --static-quantized-quals 10 --static-quantized-quals 20 --static-quantized-quals 30 --add-output-sam-program-record --create-output-bam-md5 --use-original-qualities"
],
"values": [
{
"name": "indexed_reference_fasta",
"workflow_name": "indexed_reference_fasta",
"from": "",
"type": "File",
"prefix": "-R ",
"null_flag": false,
"have_regex": false,
"default_value": [
"/home/ubuntu/references/Homo_sapiens_assembly38.fasta"
]
},
{
"name": "input_bam",
"workflow_name": "output",
"from": "SetNmMdAndUqTags",
"type": "File",
"prefix": " -I ",
"null_flag": false,
"have_regex": true,
"default_value": [
"*.tagged.sorted.md.bam"
]
},
{
"name": "input_recalibrate_bqsr",
"workflow_name": "output",
"from": "RecalibrateBQSR",
"type": "File",
"prefix": " -bqsr ",
"null_flag": false,
"have_regex": true,
"default_value": [
"*.recal_data.table"
]
},
{
"name": "output_apply_bqsr",
"workflow_name": "output_basename",
"from": "",
"type": "string",
"prefix": " -O ",
"null_flag": false,
"have_regex": false,
"default_value": [
"mother.bqsr.sorted.md.bam"
]
}
],
"outputs": [
{
"name": "output",
"datatype": "single",
"type": "File",
"partern": [
"*.bam"
],
"regex": [
"*.bam"
],
"source": ""
}
],
"parrent_task_id": [
"1-6",
"1-5"
],
"children_task_id": [
"1-9"
]
},
{
"task_id": "1-3",
"dag_id": 1,
"task_name": "#FixMate.cwl",
"task_workflow_name": "FixMate",
"docker_images": [
"vinbdi/biotools"
],
"requirement": "",
"base_command": "samtools fixmate ",
"tool_id": 0,
"parameters": [
""
],
"values": [
{
"name": "input_sam",
"workflow_name": "output",
"from": "AlignmentReads",
"type": "File",
"prefix": "",
"null_flag": false,
"have_regex": true,
"default_value": [
"*.sam"
]
},
{
"name": "output_fixmate_bam",
"workflow_name": "output_basename",
"from": "",
"type": "string",
"prefix": " -O bam ",
"null_flag": false,
"have_regex": false,
"default_value": [
"mother.lanefixed.bam"
]
}
],
"outputs": [
{
"name": "output",
"datatype": "single",
"type": "File",
"partern": [
"*.bam"
],
"regex": [
"*.bam"
],
"source": ""
}
],
"parrent_task_id": [
"1-1"
],
"children_task_id": [
"1-4"
]
},
{
"task_id": "1-4",
"dag_id": 1,
"task_name": "#MarkDuplicates.cwl",
"task_workflow_name": "MarkDuplicates",
"docker_images": [
"vinbdi/biotools"
],
"requirement": "",
"base_command": "picard MarkDuplicates ",
"tool_id": 0,
"parameters": [
" VALIDATION_STRINGENCY=SILENT OPTICAL_DUPLICATE_PIXEL_DISTANCE=2500 ASSUME_SORT_ORDER=queryname CREATE_MD5_FILE=false"
],
"values": [
{
"name": "input_bam",
"workflow_name": "output",
"from": "FixMate",
"type": "File",
"prefix": "I=",
"null_flag": false,
"have_regex": true,
"default_value": [
"*.bam"
]
},
{
"name": "output_md_bam",
"workflow_name": "output_basename",
"from": "",
"type": "string",
"prefix": " O=",
"null_flag": false,
"have_regex": false,
"default_value": [
"mother.md.bam"
]
},
{
"name": "output_report_txt",
"workflow_name": "output_basename",
"from": "",
"type": "string",
"prefix": " M=",
"null_flag": false,
"have_regex": false,
"default_value": [
"mother.bam.txt"
]
}
],
"outputs": [
{
"name": "output",
"datatype": "single",
"type": "File",
"partern": [
"*.md.bam"
],
"regex": [
"*.md.bam"
],
"source": ""
}
],
"parrent_task_id": [
"1-3"
],
"children_task_id": [
"1-7"
]
},
{
"task_id": "1-5",
"dag_id": 1,
"task_name": "#RecalibrateBQSR.cwl",
"task_workflow_name": "RecalibrateBQSR",
"docker_images": [
"vinbdi/biotools"
],
"requirement": "",
"base_command": "gatk BaseRecalibrator ",
"tool_id": 0,
"parameters": [
""
],
"values": [
{
"name": "indexed_reference_dpsnp_hg38",
"workflow_name": "indexed_reference_dpsnp_hg38",
"from": "",
"type": "File",
"prefix": " --known-sites ",
"null_flag": false,
"have_regex": false,
"default_value": [
"/home/ubuntu/references/Homo_sapiens_assembly38.dbsnp138.vcf.gz"
]
},
{
"name": "indexed_reference_fasta",
"workflow_name": "indexed_reference_fasta",
"from": "",
"type": "File",
"prefix": "-R ",
"null_flag": false,
"have_regex": false,
"default_value": [
"/home/ubuntu/references/Homo_sapiens_assembly38.fasta"
]
},
{
"name": "indexed_reference_indels_hg38",
"workflow_name": "indexed_reference_indels_hg38",
"from": "",
"type": "File",
"prefix": " --known-sites ",
"null_flag": false,
"have_regex": false,
"default_value": [
"/home/ubuntu/references/Mills_and_1000G_gold_standard.indels.hg38.vcf.gz"
]
},
{
"name": "indexed_reference_know_indels_hg38",
"workflow_name": "indexed_reference_know_indels_hg38",
"from": "",
"type": "File",
"prefix": " --known-sites ",
"null_flag": false,
"have_regex": false,
"default_value": [
"/home/ubuntu/references/Homo_sapiens_assembly38.known_indels.vcf.gz"
]
},
{
"name": "input_bam",
"workflow_name": "output",
"from": "SetNmMdAndUqTags",
"type": "File",
"prefix": " --use-original-qualities -I ",
"null_flag": false,
"have_regex": true,
"default_value": [
"*.tagged.sorted.md.bam"
]
},
{
"name": "output_recalibrate_bqsr",
"workflow_name": "output_basename",
"from": "",
"type": "string",
"prefix": " -O ",
"null_flag": false,
"have_regex": false,
"default_value": [
"mother.recal_data.table"
]
}
],
"outputs": [
{
"name": "output",
"datatype": "single",
"type": "File",
"partern": [
"*.recal_data.table"
],
"regex": [
"*.recal_data.table"
],
"source": ""
}
],
"parrent_task_id": [
"1-6"
],
"children_task_id": [
"1-2"
]
},
{
"task_id": "1-6",
"dag_id": 1,
"task_name": "#SetNmMdAndUqTags.cwl",
"task_workflow_name": "SetNmMdAndUqTags",
"docker_images": [
"vinbdi/biotools"
],
"requirement": "",
"base_command": "picard SetNmMdAndUqTags ",
"tool_id": 0,
"parameters": [
""
],
"values": [
{
"name": "indexed_reference_fasta",
"workflow_name": "indexed_reference_fasta",
"from": "",
"type": "File",
"prefix": " CREATE_INDEX=true R=",
"null_flag": false,
"have_regex": false,
"default_value": [
"/home/ubuntu/references/Homo_sapiens_assembly38.fasta"
]
},
{
"name": "input_bam",
"workflow_name": "output",
"from": "AddReplaceReadGroups",
"type": "File",
"prefix": "I=",
"null_flag": false,
"have_regex": true,
"default_value": [
"*.grp.sorted.md.bam"
]
},
{
"name": "output_add_replace_rdg",
"workflow_name": "output_basename",
"from": "",
"type": "string",
"prefix": " O=",
"null_flag": false,
"have_regex": false,
"default_value": [
"mother.tagged.sorted.md.bam"
]
}
],
"outputs": [
{
"name": "output",
"datatype": "single",
"type": "File",
"partern": [
"*.tagged.sorted.md.bam"
],
"regex": [
"*.tagged.sorted.md.bam"
],
"source": ""
}
],
"parrent_task_id": [
"1-0"
],
"children_task_id": [
"1-2",
"1-5"
]
},
{
"task_id": "1-7",
"dag_id": 1,
"task_name": "#SortBam.cwl",
"task_workflow_name": "SortBam",
"docker_images": [
"vinbdi/biotools"
],
"requirement": "",
"base_command": "picard SortSam ",
"tool_id": 0,
"parameters": [
" SO=coordinate CREATE_INDEX=false CREATE_MD5_FILE=false"
],
"values": [
{
"name": "input_bam",
"workflow_name": "output",
"from": "MarkDuplicates",
"type": "File",
"prefix": "I=",
"null_flag": false,
"have_regex": true,
"default_value": [
"*.md.bam"
]
},
{
"name": "output_sorted_bam",
"workflow_name": "output_basename",
"from": "",
"type": "string",
"prefix": " O=",
"null_flag": false,
"have_regex": false,
"default_value": [
"mother.sorted.md.bam"
]
}
],
"outputs": [
{
"name": "output",
"datatype": "single",
"type": "File",
"partern": [
"*.bam"
],
"regex": [
"*.bam"
],
"source": ""
}
],
"parrent_task_id": [
"1-4"
],
"children_task_id": [
"1-0"
]
},
{
"task_id": "1-8",
"dag_id": 1,
"task_name": "#VariantAnnotation.cwl",
"task_workflow_name": "VariantAnnotation",
"docker_images": [
"vinbdi/biotools"
],
"requirement": "",
"base_command": "vep ",
"tool_id": 0,
"parameters": [
""
],
"values": [
{
"name": "input_vcf",
"workflow_name": "output",
"from": "VariantCalling",
"type": "File",
"prefix": " -i ",
"null_flag": false,
"have_regex": true,
"default_value": [
"*.vcf.gz"
]
},
{
"name": "output_variant_annotation",
"workflow_name": "output_basename",
"from": "",
"type": "string",
"prefix": " -o ",
"null_flag": false,
"have_regex": false,
"default_value": [
"mother.vep.vcf.gz"
]
},
{
"name": "vep_cached_dir",
"workflow_name": "vep_cached_dir",
"from": "",
"type": "Directory",
"prefix": " --dir_cache ",
"null_flag": false,
"have_regex": false,
"default_value": [
"/home/ubuntu/vep"
]
},
{
"name": "vep_fasta_file",
"workflow_name": "vep_fasta_file",
"from": "",
"type": "File",
"prefix": "--format vcf -a GRCh38 --no_stats --force_overwrite --cache --offline --vcf --vcf_info_field ANN --buffer_size 60000 --phased --hgvsg --hgvs --symbol --variant_class --biotype --gene_phenotype --regulatory --ccds --transcript_version --tsl --appris --canonical --protein --uniprot --domains --sift b --polyphen b --check_existing --af --max_af --af_1kg --af_gnomad --minimal --allele_number --pubmed --fasta ",
"null_flag": false,
"have_regex": false,
"default_value": [
"/home/ubuntu/vep/homo_sapiens/101_GRCh38/Homo_sapiens.GRCh38.dna.toplevel.fa.gz"
]
}
],
"outputs": [
{
"name": "output",
"datatype": "single",
"type": "File",
"partern": [
"*.vcf.gz"
],
"regex": [
"*.vcf.gz"
],
"source": ""
}
],
"parrent_task_id": [
"1-9"
],
"children_task_id": null
},
{
"task_id": "1-9",
"dag_id": 1,
"task_name": "#VariantCalling.cwl",
"task_workflow_name": "VariantCalling",
"docker_images": [
"vinbdi/biotools"
],
"requirement": "",
"base_command": "gatk HaplotypeCaller ",
"tool_id": 0,
"parameters": [
""
],
"values": [
{
"name": "indexed_reference_fasta",
"workflow_name": "indexed_reference_fasta",
"from": "",
"type": "File",
"prefix": "-R ",
"null_flag": false,
"have_regex": false,
"default_value": [
"/home/ubuntu/references/Homo_sapiens_assembly38.fasta"
]
},
{
"name": "input_bam",
"workflow_name": "output",
"from": "ApplyBQSR",
"type": "File",
"prefix": " -I ",
"null_flag": false,
"have_regex": true,
"default_value": [
"*.bam"
]
},
{
"name": "output_variant_calling",
"workflow_name": "output_basename",
"from": "",
"type": "string",
"prefix": " -O ",
"null_flag": false,
"have_regex": false,
"default_value": [
"mother.vcf.gz"
]
}
],
"outputs": [
{
"name": "output",
"datatype": "single",
"type": "File",
"partern": [
"*.vcf.gz"
],
"regex": [
"*.vcf.gz"
],
"source": ""
}
],
"parrent_task_id": [
"1-2"
],
"children_task_id": [
"1-8"
]
}
]
}
}
{
"req": "createDag",
"files": [
{
"filename": "AddReplaceReadGroups.cwl",
"folder": "tasks",
"content": "cwlVersion: v1.0\nclass: CommandLineTool\nid: sort_bam\nrequirements:\n - class: DockerRequirement\n dockerPull: vinbdi/biotools\n - class: ResourceRequirement \n ramMin: 2000\ninputs:\n input_bam: \n type: File\n output_add_replace_rdg:\n type: string \n common_basename:\n type: string\n \noutputs:\n output: \n type: File \n outputBinding:\n glob: '*.grp.sorted.md.bam'\n secondaryFiles: [^.bai, ^.bam.md5]\narguments:\n - position: 0\n shellQuote: false \n valueFrom: >- \n I=$(inputs.input_bam) O=$(inputs.output_add_replace_rdg) CREATE_INDEX=true RGID=RG_mother RGLB=lib_mother RGPL=illumina RGPU=unit_mother RGSM=$(inputs.common_basename)\nbaseCommand: [picard, AddOrReplaceReadGroups]"
},
{
"filename": "AlignBwaMem.cwl",
"folder": "tasks",
"content": "cwlVersion: v1.0\n\nclass: CommandLineTool \nid: align_bwa_mem \n\nrequirements:\n - class: DockerRequirement\n dockerPull: vinbdi/biotools\n - class: ResourceRequirement\n ramMin: 4096\n coresMin: 4\n\ninputs:\n input_fastqs:\n type: File[]\n indexed_reference_fasta: \n type: File \n secondaryFiles: [.64.amb, .64.ann, .64.bwt, .64.pac, .64.sa, .64.alt, .dict, .fai]\n output_aligned_reads:\n type: File\n\noutputs:\n output: \n type: File \n outputBinding:\n glob: '*.sam'\n \narguments: \n - position: 0\n shellQuote: false \n valueFrom: >- \n -M -Y -t 16 -K 100000000 -p $(inputs.indexed_reference_fasta) $(inputs.input_fastqs) -o $(inputs.output_aligned_reads)\nbaseCommand: [bwa, mem]\n"
},
{
"filename": "ApplyBQSR.cwl",
"folder": "tasks",
"content": "cwlVersion: v1.0\n\nclass: CommandLineTool\nid: sort_bam\nrequirements:\n - class: DockerRequirement\n dockerPull: vinbdi/biotools\n - class: ResourceRequirement \n ramMin: 2000\n\ninputs:\n input_bam: \n type: File\n indexed_reference_fasta:\n type: File \n secondaryFiles: [.64.amb, .64.ann, .64.bwt, .64.pac, .64.sa, .64.alt, .dict, .fai]\n input_recalibrate_bqsr:\n type: File\n output_apply_bqsr:\n type: string \n \noutputs:\n output: \n type: File \n outputBinding:\n glob: '*.bam'\n secondaryFiles: [^.bai, ^.bam.md5]\n\narguments:\n - position: 0\n shellQuote: false \n valueFrom: >- \n -R $(inputs.indexed_reference_fasta) -I $(inputs.input_bam) -bqsr $(input.input_recalibrate_bqsr) -O $(inputs.output.output_apply_bqsr) --static-quantized-quals 10 --static-quantized-quals 20 --static-quantized-quals 30 --add-output-sam-program-record --create-output-bam-md5 --use-original-qualities\n\nbaseCommand: [gatk, ApplyBQSR]\n"
},
{
"filename": "FixMate.cwl",
"folder": "tasks",
"content": "cwlVersion: v1.0\n\nclass: CommandLineTool\nid: fix_mate\nrequirements:\n - class: DockerRequirement\n dockerPull: vinbdi/biotools\n - class: ResourceRequirement \n ramMin: 2000\n\ninputs:\n input_sam: \n type: File\n output_fixmate_bam:\n type: string \n \noutputs:\n output: \n type: File \n outputBinding:\n glob: '*.bam'\n secondaryFiles: [^.bai, ^.bam.md5]\n\narguments:\n - position: 0\n shellQuote: false \n valueFrom: >- \n $(inputs.input_sam) -O bam $(inputs.output_fixmate_bam)\n\nbaseCommand: [samtools, fixmate]\n"
},
{
"filename": "MarkDuplicates.cwl",
"folder": "tasks",
"content": "cwlVersion: v1.0\n\nclass: CommandLineTool\nid: fix_mate\nrequirements:\n - class: DockerRequirement\n dockerPull: vinbdi/biotools\n - class: ResourceRequirement \n ramMin: 2000\n\ninputs:\n input_bam: \n type: File\n output_md_bam:\n type: string \n output_report_txt:\n type: string\n \noutputs:\n output: \n type: File \n outputBinding:\n glob: '*.md.bam'\n secondaryFiles: [^.bai, ^.bam.md5]\n\narguments:\n - position: 0\n shellQuote: false \n valueFrom: >- \n I=$(inputs.input_bam) O=$(inputs.output_md_bam) M=$(inputs.output_report_txt) VALIDATION_STRINGENCY=SILENT OPTICAL_DUPLICATE_PIXEL_DISTANCE=2500 ASSUME_SORT_ORDER=queryname CREATE_MD5_FILE=false\n\nbaseCommand: [picard MarkDuplicates]\n"
},
{
"filename": "RecalibrateBQSR.cwl",
"folder": "tasks",
"content": "cwlVersion: v1.0\n\nclass: CommandLineTool\nid: sort_bam\nrequirements:\n - class: DockerRequirement\n dockerPull: vinbdi/biotools\n - class: ResourceRequirement \n ramMin: 2000\n\ninputs:\n input_bam: \n type: File\n indexed_reference_fasta: \n type: File \n secondaryFiles: [.64.amb, .64.ann, .64.bwt, .64.pac, .64.sa, .64.alt, .dict, .fai]\n indexed_reference_indels_hg38:\n type: File \n indexed_reference_know_indels_hg38:\n type: File \n indexed_reference_dpsnp_hg38:\n type: File\n output_recalibrate_bqsr:\n type: string \n \noutputs:\n output: \n type: File \n outputBinding:\n glob: '*.recal_data.table'\n\narguments:\n - position: 0\n shellQuote: false \n valueFrom: >- \n -R $(inputs.indexed_reference_fasta) --use-original-qualities -I $(inputs.input_bam) --known-sites $(inputs.indexed_reference_indels_hg38) --known-sites $(inputs.indexed_reference_know_indels_hg38) --known-sites $(inputs.indexed_reference_dpsnp_hg38) -O $(input.output_recalibrate_bqsr)\n\nbaseCommand: [gatk, BaseRecalibrator]\n"
},
{
"filename": "SetNmMdAndUqTags.cwl",
"folder": "tasks",
"content": "cwlVersion: v1.0\n\nclass: CommandLineTool\nid: sort_bam\nrequirements:\n - class: DockerRequirement\n dockerPull: vinbdi/biotools\n - class: ResourceRequirement \n ramMin: 2000\n\ninputs:\n input_bam: \n type: File\n indexed_reference_fasta: \n type: File \n secondaryFiles: [.64.amb, .64.ann, .64.bwt, .64.pac, .64.sa, .64.alt, .dict, .fai]\n output_add_replace_rdg:\n type: string \n \noutputs:\n output: \n type: File \n outputBinding:\n glob: '*.tagged.sorted.md.bam'\n secondaryFiles: [^.bai, ^.bam.md5]\n\narguments:\n - position: 0\n shellQuote: false \n valueFrom: >- \n I=$(inputs.input_bam) O=$(inputs.output_add_replace_rdg) CREATE_INDEX=true R=$(inputs.indexed_reference_fasta)\n\nbaseCommand: [picard, SetNmMdAndUqTags]\n"
},
{
"filename": "SortBam.cwl",
"folder": "tasks",
"content": "cwlVersion: v1.0\n\nclass: CommandLineTool\nid: sort_bam\nrequirements:\n - class: DockerRequirement\n dockerPull: vinbdi/biotools\n - class: ResourceRequirement \n ramMin: 2000\n\ninputs:\n input_bam: \n type: File\n output_sorted_bam:\n type: string \n \noutputs:\n output: \n type: File \n outputBinding:\n glob: '*.bam'\n secondaryFiles: [^.bai, ^.bam.md5]\n\narguments:\n - position: 0\n shellQuote: false \n valueFrom: >- \n I=$(inputs.input_bam) O=$(inputs.output_sorted_bam) SO=coordinate CREATE_INDEX=false CREATE_MD5_FILE=false\n\nbaseCommand: [picard, SortSam]\n"
},
{
"filename": "VariantAnnotation.cwl",
"folder": "tasks",
"content": "cwlVersion: v1.0\n\nclass: CommandLineTool\nid: sort_bam\nrequirements:\n - class: DockerRequirement\n dockerPull: vinbdi/biotools\n - class: ResourceRequirement \n ramMin: 2000\n\ninputs:\n input_vcf: \n type: File\n vep_fasta_file: \n type: File\n vep_cached_dir:\n type: Directory \n output_variant_annotation:\n type: string \n \noutputs:\n output: \n type: File \n outputBinding:\n glob: '*.vcf.gz'\n\narguments:\n - position: 0\n shellQuote: false \n valueFrom: >- \n --format vcf -a GRCh38 --no_stats --force_overwrite --cache --offline --vcf --vcf_info_field ANN --buffer_size 60000 --phased --hgvsg --hgvs --symbol --variant_class --biotype --gene_phenotype --regulatory --ccds --transcript_version --tsl --appris --canonical --protein --uniprot --domains --sift b --polyphen b --check_existing --af --max_af --af_1kg --af_gnomad --minimal --allele_number --pubmed --fasta $(inputs.vep_fasta_file) --dir_cache $(inputs.vep_cached_dir) -i $(inputs.input_vcf) -o $(inputs.output_variant_annotation)\n\nbaseCommand: [vep]\n"
},
{
"filename": "VariantCalling.cwl",
"folder": "tasks",
"content": "cwlVersion: v1.0\n\nclass: CommandLineTool\nid: sort_bam\nrequirements:\n - class: DockerRequirement\n dockerPull: vinbdi/biotools\n - class: ResourceRequirement \n ramMin: 2000\n\ninputs:\n input_bam: \n type: File\n indexed_reference_fasta:\n type: File \n secondaryFiles: [.64.amb, .64.ann, .64.bwt, .64.pac, .64.sa, .64.alt, .dict, .fai]\n output_variant_calling:\n type: string \n \noutputs:\n output: \n type: File \n outputBinding:\n glob: '*.vcf.gz'\n secondaryFiles: [^.vcf.gz.tbi]\n\narguments:\n - position: 0\n shellQuote: false \n valueFrom: >- \n -R $(inputs.indexed_reference_fasta) -I $(inputs.input_bam) -O $(input.output_variant_calling)\n\nbaseCommand: [gatk, HaplotypeCaller]\n"
},
{
"filename": "mash.cwl",
"folder": "/",
"content": "cwlVersion: v1.0\n\ndoc: | \n This workflow is used to process in DG4VN project \n\nclass: Workflow\nid: mash_workflow\nrequirements:\n - class: StepInputExpressionRequirement\n\ninputs:\n input_fastqs:\n type: File[]\n indexed_reference_fasta:\n type: File \n secondaryFiles: [.64.amb, .64.ann, .64.bwt, .64.pac, .64.sa, .64.alt, .dict, .fai]\n indexed_reference_indels_hg38:\n type: File\n indexed_reference_know_indels_hg38:\n type: File \n indexed_reference_dpsnp_hg38:\n type: File\n vep_fasta_file:\n type: File\n vep_cached_dir:\n type: Directory\n output_basename: string\n\noutputs:\n vcf_calling:\n type: File \n outputSource: VariantAnnotation/output\n\nsteps:\n AlignmentReads:\n run: tasks/AlignBwaMem.cwl \n in: \n input_fastqs: input_fastqs\n indexed_reference_fasta: indexed_reference_fasta\n output_aligned_reads:\n source: output_basename\n valueFrom: $(self + \".sam\")\n out:\n [output]\n \n FixMate:\n run: tasks/FixMate.cwl \n in: \n input_sam: AlignmentReads/output\n output_fixmate_bam: \n source: output_basename\n valueFrom: $(self + \".lanefixed.bam\")\n out:\n [output]\n \n MarkDuplicates: \n run: tasks/MarkDuplicates.cwl \n in: \n input_bam: FixMate/output\n output_md_bam:\n source: output_basename\n valueFrom: $(self + \".md.bam\")\n output_report_txt:\n source: output_basename\n valueFrom: $(self + \".bam.txt\")\n out:\n [output]\n \n SortBam: \n run: tasks/SortBam.cwl \n in: \n input_bam: MarkDuplicates/output\n output_sorted_bam:\n source: output_basename \n valueFrom: $(self + \".sorted.md.bam\")\n out: \n [output]\n\n AddReplaceReadGroups: \n run: tasks/AddReplaceReadGroups.cwl \n in: \n input_bam: SortBam/output\n output_add_replace_rdg: \n source: output_basename\n valueFrom: $(self + \".grp.sorted.md.bam\")\n common_basename: \n source: output_basename\n out:\n [output]\n\n SetNmMdAndUqTags: \n run: tasks/SetNmMdAndUqTags.cwl \n in: \n input_bam: AddReplaceReadGroups/output\n indexed_reference_fasta: indexed_reference_fasta\n output_add_replace_rdg: \n source: output_basename\n valueFrom: $(self + \".tagged.sorted.md.bam\")\n out:\n [output]\n\n RecalibrateBQSR: \n run: tasks/RecalibrateBQSR.cwl \n in: \n input_bam: SetNmMdAndUqTags/output\n indexed_reference_fasta: indexed_reference_fasta\n indexed_reference_indels_hg38: indexed_reference_indels_hg38\n indexed_reference_know_indels_hg38: indexed_reference_know_indels_hg38\n indexed_reference_dpsnp_hg38: indexed_reference_dpsnp_hg38\n output_recalibrate_bqsr:\n source: output_basename\n valueFrom: $(self + \".recal_data.table\")\n out:\n [output]\n\n ApplyBQSR: \n run: tasks/ApplyBQSR.cwl \n in: \n input_bam: SetNmMdAndUqTags/output\n input_recalibrate_bqsr: RecalibrateBQSR/output\n indexed_reference_fasta: indexed_reference_fasta\n output_apply_bqsr: \n source: output_basename\n valueFrom: $(self + \".bqsr.sorted.md.bam\")\n out:\n [output]\n\n VariantCalling: \n run: tasks/VariantCalling.cwl \n in: \n input_bam: ApplyBQSR/output\n indexed_reference_fasta: indexed_reference_fasta\n output_variant_calling: \n source: output_basename\n valueFrom: $(self + \".vcf.gz\")\n out:\n [output]\n\n VariantAnnotation: \n run: tasks/VariantAnnotation.cwl \n in: \n input_vcf: VariantCalling/output\n vep_fasta_file: vep_fasta_file\n vep_cached_dir: vep_cached_dir\n output_variant_annotation: \n source: output_basename\n valueFrom: $(self + \".vep.vcf.gz\")\n out:\n [output]\n"
},
{
"filename": "mash.yml",
"folder":"/",
"content": "input_fastqs:\n - /home/ubuntu/input/mother_R1.fq.gz\n - /home/ubuntu/input/mother_R2.fq.gz\nindexed_reference_fasta:\n class: File \n path: /home/ubuntu/references/Homo_sapiens_assembly38.fasta\nindexed_reference_indels_hg38: \n class: File \n path: /home/ubuntu/references/Mills_and_1000G_gold_standard.indels.hg38.vcf.gz\nindexed_reference_know_indels_hg38:\n class: File \n path: /home/ubuntu/references/Homo_sapiens_assembly38.known_indels.vcf.gz\nindexed_reference_dpsnp_hg38:\n class: File\n path: /home/ubuntu/references/Homo_sapiens_assembly38.dbsnp138.vcf.gz\nvep_fasta_file:\n class: File\n path: /home/ubuntu/vep/homo_sapiens/101_GRCh38/Homo_sapiens.GRCh38.dna.toplevel.fa.gz\nvep_cached_dir:\n class: Directory\n path: /home/ubuntu/vep\noutput_basename: mother\n"
}
],
"name": "mash-test-2",
"folder": "/",
"cwlfile": "mash.cwl",
"inputfile": "mash.yml",
"dagid": 1,
"userid": 1
}
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment