Last active
November 21, 2018 14:50
-
-
Save dleehr/9bd2155a6cbf5783c4f6c8d9c32d16e3 to your computer and use it in GitHub Desktop.
Updates with fixes for intervals and after meeting with GAAB
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
{ | |
"$graph": [ | |
{ | |
"class": "Workflow", | |
"inputs": [ | |
{ | |
"type": [ | |
"null", | |
{ | |
"type": "array", | |
"items": "File" | |
} | |
], | |
"id": "#exomeseq-00-prepare-reference-data.cwl/bait_intervals" | |
}, | |
{ | |
"type": "File", | |
"secondaryFiles": [ | |
".amb", | |
".ann", | |
".bwt", | |
".pac", | |
".sa", | |
".fai", | |
"^.dict" | |
], | |
"id": "#exomeseq-00-prepare-reference-data.cwl/reference_genome" | |
}, | |
{ | |
"type": [ | |
"null", | |
{ | |
"type": "array", | |
"items": "File" | |
} | |
], | |
"id": "#exomeseq-00-prepare-reference-data.cwl/target_intervals" | |
} | |
], | |
"outputs": [ | |
{ | |
"type": "File", | |
"outputSource": "#exomeseq-00-prepare-reference-data.cwl/make_bait_interval_list/output_interval_list_file", | |
"id": "#exomeseq-00-prepare-reference-data.cwl/bait_interval_list" | |
}, | |
{ | |
"type": "File", | |
"outputSource": "#exomeseq-00-prepare-reference-data.cwl/make_target_interval_list/output_interval_list_file", | |
"id": "#exomeseq-00-prepare-reference-data.cwl/target_interval_list" | |
} | |
], | |
"steps": [ | |
{ | |
"run": "#picard-BedToIntervalList.cwl", | |
"requirements": [ | |
{ | |
"class": "ResourceRequirement", | |
"coresMin": 1, | |
"ramMin": 4000, | |
"outdirMin": 12000, | |
"tmpdirMin": 12000 | |
} | |
], | |
"in": [ | |
{ | |
"source": "#exomeseq-00-prepare-reference-data.cwl/bait_intervals", | |
"id": "#exomeseq-00-prepare-reference-data.cwl/make_bait_interval_list/input_file" | |
}, | |
{ | |
"source": "#exomeseq-00-prepare-reference-data.cwl/reference_genome", | |
"id": "#exomeseq-00-prepare-reference-data.cwl/make_bait_interval_list/reference_sequence" | |
} | |
], | |
"out": [ | |
"#exomeseq-00-prepare-reference-data.cwl/make_bait_interval_list/output_interval_list_file" | |
], | |
"id": "#exomeseq-00-prepare-reference-data.cwl/make_bait_interval_list" | |
}, | |
{ | |
"run": "#picard-BedToIntervalList.cwl", | |
"requirements": [ | |
{ | |
"class": "ResourceRequirement", | |
"coresMin": 1, | |
"ramMin": 4000, | |
"outdirMin": 12000, | |
"tmpdirMin": 12000 | |
} | |
], | |
"in": [ | |
{ | |
"source": "#exomeseq-00-prepare-reference-data.cwl/target_intervals", | |
"id": "#exomeseq-00-prepare-reference-data.cwl/make_target_interval_list/input_file" | |
}, | |
{ | |
"source": "#exomeseq-00-prepare-reference-data.cwl/reference_genome", | |
"id": "#exomeseq-00-prepare-reference-data.cwl/make_target_interval_list/reference_sequence" | |
} | |
], | |
"out": [ | |
"#exomeseq-00-prepare-reference-data.cwl/make_target_interval_list/output_interval_list_file" | |
], | |
"id": "#exomeseq-00-prepare-reference-data.cwl/make_target_interval_list" | |
} | |
], | |
"id": "#exomeseq-00-prepare-reference-data.cwl" | |
}, | |
{ | |
"class": "Workflow", | |
"requirements": [ | |
{ | |
"class": "ScatterFeatureRequirement" | |
}, | |
{ | |
"class": "SchemaDefRequirement", | |
"types": [ | |
{ | |
"name": "#bespin-types.yml/FASTQReadPairType", | |
"type": "record", | |
"fields": [ | |
{ | |
"name": "#bespin-types.yml/FASTQReadPairType/name", | |
"type": "string" | |
}, | |
{ | |
"name": "#bespin-types.yml/FASTQReadPairType/read1_files", | |
"type": { | |
"type": "array", | |
"items": "File" | |
} | |
}, | |
{ | |
"name": "#bespin-types.yml/FASTQReadPairType/read2_files", | |
"type": { | |
"type": "array", | |
"items": "File" | |
} | |
} | |
] | |
}, | |
{ | |
"name": "#bespin-types.yml/ExomeseqStudyType", | |
"type": "enum", | |
"label": "The type of study", | |
"symbols": [ | |
"#bespin-types.yml/ExomeseqStudyType/Small Familial", | |
"#bespin-types.yml/ExomeseqStudyType/Large Population" | |
] | |
} | |
], | |
"id": "#bespin-types.yml", | |
"name": "#bespin-types.yml" | |
} | |
], | |
"inputs": [ | |
{ | |
"type": "File", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/bait_interval_list" | |
}, | |
{ | |
"type": [ | |
"null", | |
"int" | |
], | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/interval_padding" | |
}, | |
{ | |
"type": [ | |
"null", | |
{ | |
"type": "array", | |
"items": "File" | |
} | |
], | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/intervals" | |
}, | |
{ | |
"type": { | |
"type": "array", | |
"items": "File" | |
}, | |
"secondaryFiles": [ | |
".idx" | |
], | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/known_sites" | |
}, | |
{ | |
"type": "string", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/library" | |
}, | |
{ | |
"type": "string", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/platform" | |
}, | |
{ | |
"type": "#bespin-types.yml/FASTQReadPairType", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/read_pair" | |
}, | |
{ | |
"type": "File", | |
"secondaryFiles": [ | |
".amb", | |
".ann", | |
".bwt", | |
".pac", | |
".sa", | |
".fai", | |
"^.dict" | |
], | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/reference_genome" | |
}, | |
{ | |
"type": "File", | |
"secondaryFiles": [ | |
".idx" | |
], | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/resource_dbsnp" | |
}, | |
{ | |
"type": "File", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/target_interval_list" | |
}, | |
{ | |
"type": "int", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/threads" | |
} | |
], | |
"outputs": [ | |
{ | |
"type": { | |
"type": "array", | |
"items": "File" | |
}, | |
"outputSource": "#exomeseq-gatk4-01-preprocessing.cwl/qc/output_qc_report", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/fastqc_reports" | |
}, | |
{ | |
"type": "File", | |
"outputSource": "#exomeseq-gatk4-01-preprocessing.cwl/variant_calling/output_bam", | |
"doc": "BAM file containing assembled haplotypes and locally realigned reads", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/haplotypes_bam" | |
}, | |
{ | |
"type": "File", | |
"outputSource": "#exomeseq-gatk4-01-preprocessing.cwl/mark_duplicates/output_dedup_bam_file", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/markduplicates_bam" | |
}, | |
{ | |
"type": "File", | |
"outputSource": "#exomeseq-gatk4-01-preprocessing.cwl/variant_calling/output_variants", | |
"doc": "VCF file from per sample variant calling", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/raw_variants" | |
}, | |
{ | |
"type": "File", | |
"outputSource": "#exomeseq-gatk4-01-preprocessing.cwl/recalibrate_02_apply_bqsr/output_recalibrated_bam", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/recalibrated_reads" | |
}, | |
{ | |
"type": "File", | |
"outputSource": "#exomeseq-gatk4-01-preprocessing.cwl/recalibrate_01_analyze/output_recalibration_report", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/recalibration_table" | |
}, | |
{ | |
"type": { | |
"type": "array", | |
"items": "File" | |
}, | |
"outputSource": "#exomeseq-gatk4-01-preprocessing.cwl/trim/trim_reports", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/trim_reports" | |
} | |
], | |
"steps": [ | |
{ | |
"run": "#concat-gz-files.cwl", | |
"scatter": [ | |
"#exomeseq-gatk4-01-preprocessing.cwl/combine_reads/files", | |
"#exomeseq-gatk4-01-preprocessing.cwl/combine_reads/output_filename" | |
], | |
"scatterMethod": "dotproduct", | |
"in": [ | |
{ | |
"source": "#exomeseq-gatk4-01-preprocessing.cwl/file_pair_details/reads", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/combine_reads/files" | |
}, | |
{ | |
"source": "#exomeseq-gatk4-01-preprocessing.cwl/generate_sample_filenames/combined_reads_output_filenames", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/combine_reads/output_filename" | |
} | |
], | |
"out": [ | |
"#exomeseq-gatk4-01-preprocessing.cwl/combine_reads/output" | |
], | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/combine_reads" | |
}, | |
{ | |
"run": "#extract-named-file-pair-details.cwl", | |
"in": [ | |
{ | |
"source": "#exomeseq-gatk4-01-preprocessing.cwl/library", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/file_pair_details/library" | |
}, | |
{ | |
"source": "#exomeseq-gatk4-01-preprocessing.cwl/platform", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/file_pair_details/platform" | |
}, | |
{ | |
"source": "#exomeseq-gatk4-01-preprocessing.cwl/read_pair", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/file_pair_details/read_pair" | |
} | |
], | |
"out": [ | |
"#exomeseq-gatk4-01-preprocessing.cwl/file_pair_details/reads", | |
"#exomeseq-gatk4-01-preprocessing.cwl/file_pair_details/read_pair_name", | |
"#exomeseq-gatk4-01-preprocessing.cwl/file_pair_details/read_group_header" | |
], | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/file_pair_details" | |
}, | |
{ | |
"run": "#generate-sample-filenames.cwl", | |
"in": [ | |
{ | |
"source": "#exomeseq-gatk4-01-preprocessing.cwl/file_pair_details/read_pair_name", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/generate_sample_filenames/sample_name" | |
} | |
], | |
"out": [ | |
"#exomeseq-gatk4-01-preprocessing.cwl/generate_sample_filenames/combined_reads_output_filenames", | |
"#exomeseq-gatk4-01-preprocessing.cwl/generate_sample_filenames/mapped_reads_output_filename", | |
"#exomeseq-gatk4-01-preprocessing.cwl/generate_sample_filenames/sorted_reads_output_filename", | |
"#exomeseq-gatk4-01-preprocessing.cwl/generate_sample_filenames/dedup_reads_output_filename", | |
"#exomeseq-gatk4-01-preprocessing.cwl/generate_sample_filenames/dedup_metrics_output_filename", | |
"#exomeseq-gatk4-01-preprocessing.cwl/generate_sample_filenames/recal_reads_output_filename", | |
"#exomeseq-gatk4-01-preprocessing.cwl/generate_sample_filenames/recal_table_output_filename", | |
"#exomeseq-gatk4-01-preprocessing.cwl/generate_sample_filenames/raw_variants_output_filename", | |
"#exomeseq-gatk4-01-preprocessing.cwl/generate_sample_filenames/haplotypes_bam_output_filename" | |
], | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/generate_sample_filenames" | |
}, | |
{ | |
"run": "#gitc-bwa-mem-samtools.cwl", | |
"requirements": [ | |
{ | |
"class": "ResourceRequirement", | |
"coresMin": "$(inputs.threads)", | |
"ramMin": 16000, | |
"outdirMin": 12000, | |
"tmpdirMin": 12000 | |
} | |
], | |
"in": [ | |
{ | |
"source": "#exomeseq-gatk4-01-preprocessing.cwl/generate_sample_filenames/mapped_reads_output_filename", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/map/output_filename" | |
}, | |
{ | |
"source": "#exomeseq-gatk4-01-preprocessing.cwl/file_pair_details/read_group_header", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/map/read_group_header" | |
}, | |
{ | |
"source": "#exomeseq-gatk4-01-preprocessing.cwl/trim/trimmed_reads", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/map/reads" | |
}, | |
{ | |
"source": "#exomeseq-gatk4-01-preprocessing.cwl/reference_genome", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/map/reference" | |
}, | |
{ | |
"source": "#exomeseq-gatk4-01-preprocessing.cwl/threads", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/map/threads" | |
} | |
], | |
"out": [ | |
"#exomeseq-gatk4-01-preprocessing.cwl/map/output" | |
], | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/map" | |
}, | |
{ | |
"run": "#GATK4-MarkDuplicates.cwl", | |
"requirements": [ | |
{ | |
"class": "ResourceRequirement", | |
"ramMin": 7000, | |
"outdirMin": 12000, | |
"tmpdirMin": 12000 | |
} | |
], | |
"in": [ | |
{ | |
"default": "coordinate", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/mark_duplicates/assume_sort_order" | |
}, | |
{ | |
"default": "true", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/mark_duplicates/create_index" | |
}, | |
{ | |
"source": "#exomeseq-gatk4-01-preprocessing.cwl/sort/output_sorted_bam", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/mark_duplicates/input_file" | |
}, | |
{ | |
"default": "-Xms4000m", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/mark_duplicates/java_opt" | |
}, | |
{ | |
"source": "#exomeseq-gatk4-01-preprocessing.cwl/generate_sample_filenames/dedup_metrics_output_filename", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/mark_duplicates/metrics_filename" | |
}, | |
{ | |
"default": 2500, | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/mark_duplicates/optical_duplicate_pixel_distance" | |
}, | |
{ | |
"source": "#exomeseq-gatk4-01-preprocessing.cwl/generate_sample_filenames/dedup_reads_output_filename", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/mark_duplicates/output_filename" | |
}, | |
{ | |
"default": "SILENT", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/mark_duplicates/validation_stringency" | |
} | |
], | |
"out": [ | |
"#exomeseq-gatk4-01-preprocessing.cwl/mark_duplicates/output_dedup_bam_file", | |
"#exomeseq-gatk4-01-preprocessing.cwl/mark_duplicates/output_metrics_file" | |
], | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/mark_duplicates" | |
}, | |
{ | |
"run": "#fastqc.cwl", | |
"requirements": [ | |
{ | |
"class": "ResourceRequirement", | |
"coresMin": 4, | |
"ramMin": 2500 | |
} | |
], | |
"scatter": "#exomeseq-gatk4-01-preprocessing.cwl/qc/input_fastq_file", | |
"in": [ | |
{ | |
"source": "#exomeseq-gatk4-01-preprocessing.cwl/combine_reads/output", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/qc/input_fastq_file" | |
}, | |
{ | |
"default": 4, | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/qc/threads" | |
} | |
], | |
"out": [ | |
"#exomeseq-gatk4-01-preprocessing.cwl/qc/output_qc_report" | |
], | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/qc" | |
}, | |
{ | |
"run": "#GATK4-BaseRecalibrator.cwl", | |
"requirements": [ | |
{ | |
"class": "ResourceRequirement", | |
"ramMin": 6000 | |
} | |
], | |
"in": [ | |
{ | |
"source": "#exomeseq-gatk4-01-preprocessing.cwl/mark_duplicates/output_dedup_bam_file", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/recalibrate_01_analyze/input_bam" | |
}, | |
{ | |
"source": "#exomeseq-gatk4-01-preprocessing.cwl/interval_padding", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/recalibrate_01_analyze/interval_padding" | |
}, | |
{ | |
"source": "#exomeseq-gatk4-01-preprocessing.cwl/intervals", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/recalibrate_01_analyze/intervals" | |
}, | |
{ | |
"default": "-Xms4000m", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/recalibrate_01_analyze/java_opt" | |
}, | |
{ | |
"source": "#exomeseq-gatk4-01-preprocessing.cwl/known_sites", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/recalibrate_01_analyze/known_sites" | |
}, | |
{ | |
"source": "#exomeseq-gatk4-01-preprocessing.cwl/generate_sample_filenames/recal_table_output_filename", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/recalibrate_01_analyze/output_recalibration_report_filename" | |
}, | |
{ | |
"source": "#exomeseq-gatk4-01-preprocessing.cwl/reference_genome", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/recalibrate_01_analyze/reference" | |
} | |
], | |
"out": [ | |
"#exomeseq-gatk4-01-preprocessing.cwl/recalibrate_01_analyze/output_recalibration_report" | |
], | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/recalibrate_01_analyze" | |
}, | |
{ | |
"run": "#GATK4-ApplyBQSR.cwl", | |
"requirements": [ | |
{ | |
"class": "ResourceRequirement", | |
"ramMin": 3500 | |
} | |
], | |
"in": [ | |
{ | |
"default": true, | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/recalibrate_02_apply_bqsr/add_output_sam_program_record" | |
}, | |
{ | |
"source": "#exomeseq-gatk4-01-preprocessing.cwl/recalibrate_01_analyze/output_recalibration_report", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/recalibrate_02_apply_bqsr/bqsr_report" | |
}, | |
{ | |
"source": "#exomeseq-gatk4-01-preprocessing.cwl/mark_duplicates/output_dedup_bam_file", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/recalibrate_02_apply_bqsr/input_bam" | |
}, | |
{ | |
"source": "#exomeseq-gatk4-01-preprocessing.cwl/interval_padding", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/recalibrate_02_apply_bqsr/interval_padding" | |
}, | |
{ | |
"source": "#exomeseq-gatk4-01-preprocessing.cwl/intervals", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/recalibrate_02_apply_bqsr/intervals" | |
}, | |
{ | |
"default": "-Xms3000m", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/recalibrate_02_apply_bqsr/java_opt" | |
}, | |
{ | |
"source": "#exomeseq-gatk4-01-preprocessing.cwl/generate_sample_filenames/recal_reads_output_filename", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/recalibrate_02_apply_bqsr/output_recalibrated_bam_filename" | |
}, | |
{ | |
"source": "#exomeseq-gatk4-01-preprocessing.cwl/reference_genome", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/recalibrate_02_apply_bqsr/reference" | |
} | |
], | |
"out": [ | |
"#exomeseq-gatk4-01-preprocessing.cwl/recalibrate_02_apply_bqsr/output_recalibrated_bam" | |
], | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/recalibrate_02_apply_bqsr" | |
}, | |
{ | |
"run": "#GATK4-SortSam.cwl", | |
"requirements": [ | |
{ | |
"class": "ResourceRequirement", | |
"ramMin": 5000 | |
} | |
], | |
"in": [ | |
{ | |
"source": "#exomeseq-gatk4-01-preprocessing.cwl/map/output", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/sort/input_file" | |
}, | |
{ | |
"default": "-Xms4000m", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/sort/java_opt" | |
}, | |
{ | |
"source": "#exomeseq-gatk4-01-preprocessing.cwl/generate_sample_filenames/sorted_reads_output_filename", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/sort/output_sorted_bam_filename" | |
}, | |
{ | |
"default": "coordinate", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/sort/sort_order" | |
} | |
], | |
"out": [ | |
"#exomeseq-gatk4-01-preprocessing.cwl/sort/output_sorted_bam" | |
], | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/sort" | |
}, | |
{ | |
"run": "#trim_galore.cwl", | |
"requirements": [ | |
{ | |
"class": "ResourceRequirement", | |
"coresMin": 4, | |
"ramMin": 8000 | |
} | |
], | |
"in": [ | |
{ | |
"default": true, | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/trim/paired" | |
}, | |
{ | |
"source": "#exomeseq-gatk4-01-preprocessing.cwl/combine_reads/output", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/trim/reads" | |
} | |
], | |
"out": [ | |
"#exomeseq-gatk4-01-preprocessing.cwl/trim/trimmed_reads", | |
"#exomeseq-gatk4-01-preprocessing.cwl/trim/trim_reports" | |
], | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/trim" | |
}, | |
{ | |
"run": "#GATK4-HaplotypeCaller.cwl", | |
"requirements": [ | |
{ | |
"class": "ResourceRequirement", | |
"coresMin": 1, | |
"ramMin": 8192 | |
} | |
], | |
"in": [ | |
{ | |
"default": [ | |
"StandardAnnotation", | |
"AS_StandardAnnotation" | |
], | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/variant_calling/annotation_groups" | |
}, | |
{ | |
"source": "#exomeseq-gatk4-01-preprocessing.cwl/resource_dbsnp", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/variant_calling/dbnsp" | |
}, | |
{ | |
"default": "GVCF", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/variant_calling/emit_ref_confidence" | |
}, | |
{ | |
"source": "#exomeseq-gatk4-01-preprocessing.cwl/recalibrate_02_apply_bqsr/output_recalibrated_bam", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/variant_calling/input_bam" | |
}, | |
{ | |
"source": "#exomeseq-gatk4-01-preprocessing.cwl/interval_padding", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/variant_calling/interval_padding" | |
}, | |
{ | |
"source": "#exomeseq-gatk4-01-preprocessing.cwl/intervals", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/variant_calling/intervals" | |
}, | |
{ | |
"default": "-Xms7000m", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/variant_calling/java_opt" | |
}, | |
{ | |
"source": "#exomeseq-gatk4-01-preprocessing.cwl/generate_sample_filenames/haplotypes_bam_output_filename", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/variant_calling/output_bam_filename" | |
}, | |
{ | |
"source": "#exomeseq-gatk4-01-preprocessing.cwl/generate_sample_filenames/raw_variants_output_filename", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/variant_calling/output_variants_filename" | |
}, | |
{ | |
"source": "#exomeseq-gatk4-01-preprocessing.cwl/reference_genome", | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/variant_calling/reference" | |
} | |
], | |
"out": [ | |
"#exomeseq-gatk4-01-preprocessing.cwl/variant_calling/output_variants", | |
"#exomeseq-gatk4-01-preprocessing.cwl/variant_calling/output_bam" | |
], | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl/variant_calling" | |
} | |
], | |
"id": "#exomeseq-gatk4-01-preprocessing.cwl" | |
}, | |
{ | |
"class": "CommandLineTool", | |
"requirements": [ | |
{ | |
"class": "DockerRequirement", | |
"dockerPull": "broadinstitute/gatk:4.0.4.0" | |
}, | |
{ | |
"class": "InlineJavascriptRequirement" | |
} | |
], | |
"hints": [ | |
{ | |
"class": "SoftwareRequirement", | |
"packages": [ | |
{ | |
"version": [ | |
"4.0.4.0" | |
], | |
"https://schema.org/citation": "https://doi.org/10.1101/gr.107524.110", | |
"package": "gatk" | |
} | |
] | |
} | |
], | |
"baseCommand": "gatk", | |
"inputs": [ | |
{ | |
"type": "boolean", | |
"default": false, | |
"inputBinding": { | |
"prefix": "--add-output-sam-program-record" | |
}, | |
"id": "#GATK4-ApplyBQSR.cwl/add_output_sam_program_record" | |
}, | |
{ | |
"type": "File", | |
"doc": "BQSR recalibration report", | |
"inputBinding": { | |
"prefix": "-bqsr" | |
}, | |
"id": "#GATK4-ApplyBQSR.cwl/bqsr_report" | |
}, | |
{ | |
"type": "File", | |
"inputBinding": { | |
"prefix": "-I" | |
}, | |
"secondaryFiles": [ | |
"^.bai" | |
], | |
"doc": "input bam", | |
"id": "#GATK4-ApplyBQSR.cwl/input_bam" | |
}, | |
{ | |
"type": [ | |
"null", | |
"int" | |
], | |
"doc": "Amount of padding (in bp) to add to each interval you are including. Defaults to 0", | |
"inputBinding": { | |
"prefix": "-ip" | |
}, | |
"id": "#GATK4-ApplyBQSR.cwl/interval_padding" | |
}, | |
{ | |
"type": [ | |
"null", | |
{ | |
"type": "array", | |
"items": "File", | |
"inputBinding": { | |
"prefix": "-L" | |
} | |
} | |
], | |
"doc": "One or more genomic intervals over which to operate", | |
"id": "#GATK4-ApplyBQSR.cwl/intervals" | |
}, | |
{ | |
"type": "string", | |
"doc": "String of options to pass to JVM at runtime", | |
"inputBinding": { | |
"prefix": "--java-options", | |
"position": -1, | |
"shellQuote": true | |
}, | |
"id": "#GATK4-ApplyBQSR.cwl/java_opt" | |
}, | |
{ | |
"type": "string", | |
"doc": "Output BAM filename", | |
"default": "recalibrated.bam", | |
"inputBinding": { | |
"prefix": "-O" | |
}, | |
"id": "#GATK4-ApplyBQSR.cwl/output_recalibrated_bam_filename" | |
}, | |
{ | |
"type": "File", | |
"inputBinding": { | |
"position": 2, | |
"prefix": "-R" | |
}, | |
"secondaryFiles": [ | |
".amb", | |
".ann", | |
".bwt", | |
".pac", | |
".sa", | |
".fai", | |
"^.dict" | |
], | |
"id": "#GATK4-ApplyBQSR.cwl/reference" | |
}, | |
{ | |
"type": [ | |
"null", | |
{ | |
"type": "array", | |
"items": "int", | |
"inputBinding": { | |
"prefix": "--static-quantized-quals" | |
} | |
} | |
], | |
"doc": "Use static quantized quality scores to a given number of levels (with -bqsr)", | |
"id": "#GATK4-ApplyBQSR.cwl/static_quantized_quals" | |
}, | |
{ | |
"type": "boolean", | |
"default": false, | |
"inputBinding": { | |
"prefix": "--use-original-qualities" | |
}, | |
"id": "#GATK4-ApplyBQSR.cwl/use_original_qualities" | |
} | |
], | |
"outputs": [ | |
{ | |
"type": "File", | |
"outputBinding": { | |
"glob": "$(inputs.output_recalibrated_bam_filename)" | |
}, | |
"secondaryFiles": [ | |
"^.bai" | |
], | |
"id": "#GATK4-ApplyBQSR.cwl/output_recalibrated_bam" | |
} | |
], | |
"arguments": [ | |
{ | |
"valueFrom": "ApplyBQSR", | |
"position": 0 | |
} | |
], | |
"id": "#GATK4-ApplyBQSR.cwl" | |
}, | |
{ | |
"class": "CommandLineTool", | |
"requirements": [ | |
{ | |
"class": "DockerRequirement", | |
"dockerPull": "broadinstitute/gatk:4.0.4.0" | |
}, | |
{ | |
"class": "InlineJavascriptRequirement" | |
} | |
], | |
"hints": [ | |
{ | |
"class": "SoftwareRequirement", | |
"packages": [ | |
{ | |
"version": [ | |
"4.0.4.0" | |
], | |
"https://schema.org/citation": "https://doi.org/10.1101/gr.107524.110", | |
"package": "gatk" | |
} | |
] | |
} | |
], | |
"baseCommand": "gatk", | |
"inputs": [ | |
{ | |
"type": "File", | |
"inputBinding": { | |
"prefix": "-I" | |
}, | |
"secondaryFiles": [ | |
"^.bai" | |
], | |
"doc": "input bam", | |
"id": "#GATK4-BaseRecalibrator.cwl/input_bam" | |
}, | |
{ | |
"type": [ | |
"null", | |
"int" | |
], | |
"doc": "Amount of padding (in bp) to add to each interval you are including. Defaults to 0", | |
"inputBinding": { | |
"prefix": "-ip" | |
}, | |
"id": "#GATK4-BaseRecalibrator.cwl/interval_padding" | |
}, | |
{ | |
"type": [ | |
"null", | |
{ | |
"type": "array", | |
"items": "File", | |
"inputBinding": { | |
"prefix": "-L" | |
} | |
} | |
], | |
"doc": "One or more genomic intervals over which to operate", | |
"id": "#GATK4-BaseRecalibrator.cwl/intervals" | |
}, | |
{ | |
"type": "string", | |
"doc": "String of options to pass to JVM at runtime", | |
"inputBinding": { | |
"prefix": "--java-options", | |
"position": -1, | |
"shellQuote": true | |
}, | |
"id": "#GATK4-BaseRecalibrator.cwl/java_opt" | |
}, | |
{ | |
"type": [ | |
"null", | |
{ | |
"type": "array", | |
"items": "File", | |
"inputBinding": { | |
"prefix": "--known-sites" | |
} | |
} | |
], | |
"secondaryFiles": [ | |
".idx" | |
], | |
"doc": "Any number of VCF files representing known SNPs and/or indels. Could be e.g. dbSNP and/or official 1000 Genomes indel calls. SNPs in these files will be ignored unless the --mismatchFraction argument is used. optional parameter.", | |
"id": "#GATK4-BaseRecalibrator.cwl/known_sites" | |
}, | |
{ | |
"type": "string", | |
"doc": "Recalibration Report Filename", | |
"default": "recal.csv", | |
"inputBinding": { | |
"prefix": "-O" | |
}, | |
"id": "#GATK4-BaseRecalibrator.cwl/output_recalibration_report_filename" | |
}, | |
{ | |
"type": "File", | |
"inputBinding": { | |
"position": 2, | |
"prefix": "-R" | |
}, | |
"secondaryFiles": [ | |
".amb", | |
".ann", | |
".bwt", | |
".pac", | |
".sa", | |
".fai", | |
"^.dict" | |
], | |
"id": "#GATK4-BaseRecalibrator.cwl/reference" | |
}, | |
{ | |
"type": "boolean", | |
"default": false, | |
"inputBinding": { | |
"prefix": "--use-original-qualities" | |
}, | |
"id": "#GATK4-BaseRecalibrator.cwl/use_original_qualities" | |
} | |
], | |
"outputs": [ | |
{ | |
"type": "File", | |
"outputBinding": { | |
"glob": "$(inputs.output_recalibration_report_filename)" | |
}, | |
"id": "#GATK4-BaseRecalibrator.cwl/output_recalibration_report" | |
} | |
], | |
"arguments": [ | |
{ | |
"valueFrom": "BaseRecalibrator", | |
"position": 0 | |
} | |
], | |
"id": "#GATK4-BaseRecalibrator.cwl" | |
}, | |
{ | |
"class": "CommandLineTool", | |
"requirements": [ | |
{ | |
"class": "DockerRequirement", | |
"dockerPull": "broadinstitute/gatk:4.0.4.0" | |
}, | |
{ | |
"class": "InlineJavascriptRequirement" | |
} | |
], | |
"hints": [ | |
{ | |
"class": "SoftwareRequirement", | |
"packages": [ | |
{ | |
"version": [ | |
"4.0.4.0" | |
], | |
"https://schema.org/citation": "https://doi.org/10.1101/gr.107524.110", | |
"package": "gatk" | |
} | |
] | |
} | |
], | |
"baseCommand": "gatk", | |
"inputs": [ | |
{ | |
"type": [ | |
"null", | |
{ | |
"type": "array", | |
"items": "string", | |
"inputBinding": { | |
"prefix": "-G" | |
} | |
} | |
], | |
"doc": "One or more groups of annotations to apply to variant calls This argument may be specified 0 or more times. Default value: [StandardAnnotation, StandardHCAnnotation].", | |
"id": "#GATK4-HaplotypeCaller.cwl/annotation_groups" | |
}, | |
{ | |
"type": [ | |
"null", | |
"File" | |
], | |
"inputBinding": { | |
"prefix": "-D" | |
}, | |
"secondaryFiles": [ | |
".idx" | |
], | |
"doc": "dbSNP file. Default value: null", | |
"id": "#GATK4-HaplotypeCaller.cwl/dbsnp" | |
}, | |
{ | |
"type": [ | |
"null", | |
"string" | |
], | |
"inputBinding": { | |
"prefix": "-ERC" | |
}, | |
"doc": "Mode for emitting reference confidence scores Default value: NONE. Possible values: {NONE, BP_RESOLUTION, GVCF}", | |
"id": "#GATK4-HaplotypeCaller.cwl/emit_ref_confidence" | |
}, | |
{ | |
"type": "File", | |
"inputBinding": { | |
"prefix": "-I" | |
}, | |
"secondaryFiles": [ | |
"^.bai" | |
], | |
"doc": "BAM/SAM/CRAM file containing reads. Required.", | |
"id": "#GATK4-HaplotypeCaller.cwl/input_bam" | |
}, | |
{ | |
"type": [ | |
"null", | |
"int" | |
], | |
"doc": "Amount of padding (in bp) to add to each interval you are including. Defaults to 0", | |
"inputBinding": { | |
"prefix": "-ip" | |
}, | |
"id": "#GATK4-HaplotypeCaller.cwl/interval_padding" | |
}, | |
{ | |
"type": [ | |
"null", | |
{ | |
"type": "array", | |
"items": "File", | |
"inputBinding": { | |
"prefix": "-L" | |
} | |
} | |
], | |
"doc": "One or more genomic intervals over which to operate", | |
"id": "#GATK4-HaplotypeCaller.cwl/intervals" | |
}, | |
{ | |
"type": "string", | |
"doc": "String of options to pass to JVM at runtime", | |
"inputBinding": { | |
"prefix": "--java-options", | |
"position": -1, | |
"shellQuote": true | |
}, | |
"id": "#GATK4-HaplotypeCaller.cwl/java_opt" | |
}, | |
{ | |
"type": [ | |
"null", | |
"string" | |
], | |
"inputBinding": { | |
"prefix": "-bamout" | |
}, | |
"doc": "File to which assembled haplotypes should be written.", | |
"id": "#GATK4-HaplotypeCaller.cwl/output_bam_filename" | |
}, | |
{ | |
"type": "string", | |
"inputBinding": { | |
"prefix": "-O" | |
}, | |
"doc": "File to which variants should be written Required. Use .g.vcf to make a GVCF file.", | |
"id": "#GATK4-HaplotypeCaller.cwl/output_variants_filename" | |
}, | |
{ | |
"type": "File", | |
"inputBinding": { | |
"prefix": "-R" | |
}, | |
"secondaryFiles": [ | |
".amb", | |
".ann", | |
".bwt", | |
".pac", | |
".sa", | |
".fai", | |
"^.dict" | |
], | |
"doc": "Reference sequence file Required.", | |
"id": "#GATK4-HaplotypeCaller.cwl/reference" | |
} | |
], | |
"outputs": [ | |
{ | |
"type": "File", | |
"outputBinding": { | |
"glob": "$(inputs.output_bam_filename)" | |
}, | |
"secondaryFiles": [ | |
"^.bai" | |
], | |
"id": "#GATK4-HaplotypeCaller.cwl/output_bam" | |
}, | |
{ | |
"type": "File", | |
"outputBinding": { | |
"glob": "$(inputs.output_variants_filename)" | |
}, | |
"secondaryFiles": [ | |
".idx" | |
], | |
"id": "#GATK4-HaplotypeCaller.cwl/output_variants" | |
} | |
], | |
"arguments": [ | |
{ | |
"valueFrom": "HaplotypeCaller", | |
"position": 0 | |
} | |
], | |
"id": "#GATK4-HaplotypeCaller.cwl" | |
}, | |
{ | |
"class": "CommandLineTool", | |
"requirements": [ | |
{ | |
"class": "DockerRequirement", | |
"dockerPull": "broadinstitute/gatk:4.0.4.0" | |
}, | |
{ | |
"class": "InlineJavascriptRequirement" | |
} | |
], | |
"hints": [ | |
{ | |
"class": "SoftwareRequirement", | |
"packages": [ | |
{ | |
"version": [ | |
"4.0.4.0" | |
], | |
"https://schema.org/citation": "https://doi.org/10.1101/gr.107524.110", | |
"package": "gatk" | |
} | |
] | |
} | |
], | |
"baseCommand": "gatk", | |
"inputs": [ | |
{ | |
"type": [ | |
"null", | |
"string" | |
], | |
"inputBinding": { | |
"prefix": "--ASSUME_SORT_ORDER" | |
}, | |
"id": "#GATK4-MarkDuplicates.cwl/assume_sort_order" | |
}, | |
{ | |
"type": "string", | |
"default": "false", | |
"inputBinding": { | |
"prefix": "--CREATE_INDEX" | |
}, | |
"doc": "Create an index", | |
"id": "#GATK4-MarkDuplicates.cwl/create_index" | |
}, | |
{ | |
"type": "File", | |
"doc": "One or more input SAM or BAM files to analyze. Must be coordinate sorted.", | |
"inputBinding": { | |
"prefix": "--INPUT" | |
}, | |
"id": "#GATK4-MarkDuplicates.cwl/input_file" | |
}, | |
{ | |
"type": [ | |
"null", | |
"string" | |
], | |
"doc": "String of options to pass to JVM at runtime", | |
"inputBinding": { | |
"prefix": "--java-options", | |
"position": -1, | |
"shellQuote": true | |
}, | |
"id": "#GATK4-MarkDuplicates.cwl/java_opt" | |
}, | |
{ | |
"type": "string", | |
"doc": "Output metrics filename", | |
"default": "marked_dup_metrics.txt", | |
"inputBinding": { | |
"prefix": "--METRICS_FILE" | |
}, | |
"id": "#GATK4-MarkDuplicates.cwl/metrics_filename" | |
}, | |
{ | |
"type": [ | |
"null", | |
"int" | |
], | |
"inputBinding": { | |
"prefix": "--OPTICAL_DUPLICATE_PIXEL_DISTANCE" | |
}, | |
"id": "#GATK4-MarkDuplicates.cwl/optical_duplicate_pixel_distance" | |
}, | |
{ | |
"type": "string", | |
"doc": "Output filename", | |
"default": "marked_duplicates.bam", | |
"inputBinding": { | |
"prefix": "--OUTPUT" | |
}, | |
"id": "#GATK4-MarkDuplicates.cwl/output_filename" | |
}, | |
{ | |
"type": [ | |
"null", | |
"string" | |
], | |
"doc": "If true do not write duplicates to the output file instead of writing them with appropriate flags set. (Default true).", | |
"inputBinding": { | |
"prefix": "--REMOVE_DUPLICATES:", | |
"shellQuote": false | |
}, | |
"id": "#GATK4-MarkDuplicates.cwl/remove_duplicates" | |
}, | |
{ | |
"type": [ | |
"null", | |
"string" | |
], | |
"inputBinding": { | |
"prefix": "--VALIDATION_STRINGENCY" | |
}, | |
"id": "#GATK4-MarkDuplicates.cwl/validation_stringency" | |
} | |
], | |
"outputs": [ | |
{ | |
"type": "File", | |
"outputBinding": { | |
"glob": "$(inputs.output_filename)" | |
}, | |
"secondaryFiles": [ | |
"^.bai" | |
], | |
"id": "#GATK4-MarkDuplicates.cwl/output_dedup_bam_file" | |
}, | |
{ | |
"type": "File", | |
"outputBinding": { | |
"glob": "$(inputs.metrics_filename)" | |
}, | |
"id": "#GATK4-MarkDuplicates.cwl/output_metrics_file" | |
} | |
], | |
"arguments": [ | |
{ | |
"valueFrom": "MarkDuplicates", | |
"position": 0 | |
} | |
], | |
"id": "#GATK4-MarkDuplicates.cwl" | |
}, | |
{ | |
"class": "CommandLineTool", | |
"requirements": [ | |
{ | |
"class": "DockerRequirement", | |
"dockerPull": "broadinstitute/gatk:4.0.4.0" | |
}, | |
{ | |
"class": "InlineJavascriptRequirement" | |
} | |
], | |
"hints": [ | |
{ | |
"class": "SoftwareRequirement", | |
"packages": [ | |
{ | |
"version": [ | |
"4.0.4.0" | |
], | |
"https://schema.org/citation": "https://doi.org/10.1101/gr.107524.110", | |
"package": "gatk" | |
} | |
] | |
} | |
], | |
"baseCommand": "gatk", | |
"inputs": [ | |
{ | |
"type": "string", | |
"default": "false", | |
"inputBinding": { | |
"prefix": "--CREATE_INDEX" | |
}, | |
"doc": "Create an index", | |
"id": "#GATK4-SortSam.cwl/create_index" | |
}, | |
{ | |
"type": "File", | |
"doc": "One or more input SAM or BAM files to sort", | |
"inputBinding": { | |
"prefix": "--INPUT" | |
}, | |
"id": "#GATK4-SortSam.cwl/input_file" | |
}, | |
{ | |
"type": [ | |
"null", | |
"string" | |
], | |
"doc": "String of options to pass to JVM at runtime", | |
"inputBinding": { | |
"prefix": "--java-options", | |
"position": -1, | |
"shellQuote": true | |
}, | |
"id": "#GATK4-SortSam.cwl/java_opt" | |
}, | |
{ | |
"type": "string", | |
"doc": "Output filename", | |
"default": "sorted.bam", | |
"inputBinding": { | |
"prefix": "--OUTPUT" | |
}, | |
"id": "#GATK4-SortSam.cwl/output_sorted_bam_filename" | |
}, | |
{ | |
"type": "string", | |
"doc": "Sort order (e.g. coordinate)", | |
"inputBinding": { | |
"prefix": "--SORT_ORDER" | |
}, | |
"id": "#GATK4-SortSam.cwl/sort_order" | |
} | |
], | |
"outputs": [ | |
{ | |
"type": "File", | |
"outputBinding": { | |
"glob": "$(inputs.output_sorted_bam_filename)" | |
}, | |
"secondaryFiles": [ | |
"^.bai" | |
], | |
"id": "#GATK4-SortSam.cwl/output_sorted_bam" | |
} | |
], | |
"arguments": [ | |
{ | |
"valueFrom": "SortSam", | |
"position": 0 | |
} | |
], | |
"id": "#GATK4-SortSam.cwl" | |
}, | |
{ | |
"class": "CommandLineTool", | |
"requirements": [ | |
{ | |
"class": "InitialWorkDirRequirement", | |
"listing": [ | |
{ | |
"entryname": "combinefiles.sh", | |
"entry": "zcat \"$@\" | gzip\n" | |
} | |
] | |
} | |
], | |
"baseCommand": "bash", | |
"arguments": [ | |
"combinefiles.sh" | |
], | |
"inputs": [ | |
{ | |
"type": { | |
"type": "array", | |
"items": "File" | |
}, | |
"inputBinding": { | |
"position": 2, | |
"shellQuote": false | |
}, | |
"id": "#concat-gz-files.cwl/files" | |
}, | |
{ | |
"type": "string", | |
"id": "#concat-gz-files.cwl/output_filename" | |
} | |
], | |
"stdout": "$(inputs.output_filename)", | |
"outputs": [ | |
{ | |
"type": "File", | |
"outputBinding": { | |
"glob": "$(inputs.output_filename)" | |
}, | |
"id": "#concat-gz-files.cwl/output" | |
} | |
], | |
"id": "#concat-gz-files.cwl" | |
}, | |
{ | |
"class": "ExpressionTool", | |
"label": "Given a FASTQReadPairType returns a 2D array of the files contained within", | |
"requirements": [ | |
{ | |
"class": "InlineJavascriptRequirement" | |
}, | |
{ | |
"$import": "#bespin-types.yml" | |
} | |
], | |
"inputs": [ | |
{ | |
"type": "string", | |
"id": "#extract-named-file-pair-details.cwl/library" | |
}, | |
{ | |
"type": "string", | |
"id": "#extract-named-file-pair-details.cwl/platform" | |
}, | |
{ | |
"type": "#bespin-types.yml/FASTQReadPairType", | |
"id": "#extract-named-file-pair-details.cwl/read_pair" | |
} | |
], | |
"outputs": [ | |
{ | |
"type": "string", | |
"id": "#extract-named-file-pair-details.cwl/read_group_header" | |
}, | |
{ | |
"type": "string", | |
"id": "#extract-named-file-pair-details.cwl/read_pair_name" | |
}, | |
{ | |
"type": { | |
"type": "array", | |
"items": { | |
"type": "array", | |
"items": "File" | |
} | |
}, | |
"id": "#extract-named-file-pair-details.cwl/reads" | |
} | |
], | |
"expression": "${\n var readPairName = inputs.read_pair.name;\n var readGroupHeader = \"@RG\" +\n \"\\\\tID:\" + readPairName +\n \"\\\\tLB:\" + inputs.library +\n \"\\\\tPL:\" + inputs.platform +\n \"\\\\tPU:\" + readPairName +\n \"\\\\tSM:\" + readPairName;\n return {\n reads: [\n inputs.read_pair.read1_files,\n inputs.read_pair.read2_files,\n ],\n read_pair_name: readPairName,\n read_group_header: readGroupHeader\n };\n\n}\n", | |
"id": "#extract-named-file-pair-details.cwl" | |
}, | |
{ | |
"class": "CommandLineTool", | |
"hints": [ | |
{ | |
"class": "DockerRequirement", | |
"dockerPull": "dukegcb/fastqc:0.11.4" | |
}, | |
{ | |
"class": "SoftwareRequirement", | |
"packages": [ | |
{ | |
"version": [ | |
"0.11.4" | |
], | |
"package": "fastqc", | |
"https://schema.org/citation": "https://www.bioinformatics.babraham.ac.uk/projects/fastqc/" | |
} | |
] | |
} | |
], | |
"requirements": [ | |
{ | |
"class": "InlineJavascriptRequirement" | |
} | |
], | |
"inputs": [ | |
{ | |
"type": "string", | |
"default": "fastq", | |
"inputBinding": { | |
"position": 3, | |
"prefix": "--format" | |
}, | |
"id": "#fastqc.cwl/format" | |
}, | |
{ | |
"type": "File", | |
"inputBinding": { | |
"position": 4 | |
}, | |
"id": "#fastqc.cwl/input_fastq_file" | |
}, | |
{ | |
"type": "boolean", | |
"default": true, | |
"inputBinding": { | |
"prefix": "--noextract", | |
"position": 2 | |
}, | |
"id": "#fastqc.cwl/noextract" | |
}, | |
{ | |
"type": "int", | |
"default": 1, | |
"doc": "The number of threads to use. The fastqc wrapper script requests 250MB of memory per thread, so keep this in line with any CWL ResourceRequirement", | |
"inputBinding": { | |
"position": 5, | |
"prefix": "--threads" | |
}, | |
"id": "#fastqc.cwl/threads" | |
} | |
], | |
"outputs": [ | |
{ | |
"type": "File", | |
"outputBinding": { | |
"glob": "*_fastqc.zip" | |
}, | |
"id": "#fastqc.cwl/output_qc_report" | |
} | |
], | |
"baseCommand": "fastqc", | |
"arguments": [ | |
{ | |
"valueFrom": "$('/tmp')", | |
"prefix": "--dir", | |
"position": 5 | |
}, | |
{ | |
"valueFrom": "$(runtime.outdir)", | |
"prefix": "-o", | |
"position": 5 | |
} | |
], | |
"id": "#fastqc.cwl" | |
}, | |
{ | |
"class": "ExpressionTool", | |
"label": "Generates a set of file names for preprocessing steps based on an input sample name", | |
"requirements": [ | |
{ | |
"class": "InlineJavascriptRequirement" | |
} | |
], | |
"inputs": [ | |
{ | |
"type": "string", | |
"id": "#generate-sample-filenames.cwl/sample_name" | |
} | |
], | |
"outputs": [ | |
{ | |
"type": { | |
"type": "array", | |
"items": "string" | |
}, | |
"id": "#generate-sample-filenames.cwl/combined_reads_output_filenames" | |
}, | |
{ | |
"type": "string", | |
"id": "#generate-sample-filenames.cwl/dedup_metrics_output_filename" | |
}, | |
{ | |
"type": "string", | |
"id": "#generate-sample-filenames.cwl/dedup_reads_output_filename" | |
}, | |
{ | |
"type": "string", | |
"id": "#generate-sample-filenames.cwl/fixedtag_reads_output_filename" | |
}, | |
{ | |
"type": "string", | |
"id": "#generate-sample-filenames.cwl/haplotypes_bam_output_filename" | |
}, | |
{ | |
"type": "string", | |
"id": "#generate-sample-filenames.cwl/hs_metrics_output_filename" | |
}, | |
{ | |
"type": "string", | |
"id": "#generate-sample-filenames.cwl/mapped_reads_output_filename" | |
}, | |
{ | |
"type": "string", | |
"id": "#generate-sample-filenames.cwl/raw_variants_output_filename" | |
}, | |
{ | |
"type": "string", | |
"id": "#generate-sample-filenames.cwl/recal_reads_output_filename" | |
}, | |
{ | |
"type": "string", | |
"id": "#generate-sample-filenames.cwl/recal_table_output_filename" | |
}, | |
{ | |
"type": "string", | |
"id": "#generate-sample-filenames.cwl/sorted_reads_output_filename" | |
} | |
], | |
"expression": "${\n function makeFilename(base, suffix, extension) {\n return base + '-' + suffix + '.' + extension;\n }\n var base = inputs.sample_name\n\n return {\n combined_reads_output_filenames: [\n makeFilename(base, 'R1', 'fastq.gz'),\n makeFilename(base, 'R2', 'fastq.gz'),\n ],\n mapped_reads_output_filename: makeFilename(base, 'mapped', 'bam'),\n sorted_reads_output_filename: makeFilename(base, 'sorted', 'bam'),\n dedup_reads_output_filename: makeFilename(base, 'dedup', 'bam'),\n dedup_metrics_output_filename: makeFilename(base, 'dedup-metrics', 'out'),\n recal_reads_output_filename: makeFilename(base, 'recal', 'bam'),\n recal_table_output_filename: makeFilename(base, 'recal', 'table'),\n raw_variants_output_filename: makeFilename(base, 'raw_variants', 'g.vcf'),\n haplotypes_bam_output_filename: makeFilename(base, 'haplotypes', 'bam'),\n hs_metrics_output_filename: makeFilename(base, 'hs', 'txt'),\n fixedtag_reads_output_filename: makeFilename(base, 'fixed_tags', 'bam')\n };\n}\n", | |
"id": "#generate-sample-filenames.cwl" | |
}, | |
{ | |
"class": "CommandLineTool", | |
"requirements": [ | |
{ | |
"class": "DockerRequirement", | |
"dockerPull": "broadinstitute/genomes-in-the-cloud:2.3.1-1512499786" | |
}, | |
{ | |
"class": "InlineJavascriptRequirement" | |
}, | |
{ | |
"class": "InitialWorkDirRequirement", | |
"listing": [ | |
{ | |
"entryname": "bwa-mem-samtools.sh", | |
"entry": "/usr/gitc/bwa mem $@ | samtools view -1 -\n" | |
} | |
] | |
} | |
], | |
"baseCommand": "bash", | |
"arguments": [ | |
"bwa-mem-samtools.sh" | |
], | |
"hints": [ | |
{ | |
"class": "SoftwareRequirement", | |
"packages": [ | |
{ | |
"version": [ | |
"0.7.15" | |
], | |
"package": "bwa", | |
"https://schema.org/citation": "https://dx.doi.org/10.1093/bioinformatics/btp324" | |
}, | |
{ | |
"version": [ | |
"1.3.1" | |
], | |
"package": "samtools", | |
"https://schema.org/citation": "https://doi.org/10.1093/bioinformatics/btp352" | |
} | |
] | |
} | |
], | |
"inputs": [ | |
{ | |
"type": [ | |
"null", | |
{ | |
"type": "array", | |
"items": "int" | |
} | |
], | |
"inputBinding": { | |
"position": 1, | |
"prefix": "-I", | |
"itemSeparator": "," | |
}, | |
"id": "#gitc-bwa-mem-samtools.cwl/min_std_max_min" | |
}, | |
{ | |
"type": [ | |
"null", | |
"int" | |
], | |
"inputBinding": { | |
"position": 1, | |
"prefix": "-k" | |
}, | |
"doc": "-k INT minimum seed length [19]", | |
"id": "#gitc-bwa-mem-samtools.cwl/minimum_seed_length" | |
}, | |
{ | |
"type": "string", | |
"id": "#gitc-bwa-mem-samtools.cwl/output_filename" | |
}, | |
{ | |
"type": [ | |
"null", | |
"string" | |
], | |
"inputBinding": { | |
"position": 1, | |
"prefix": "-R" | |
}, | |
"doc": "-R STR read group header line such as '@RG\\tID:foo\\tSM:bar' [null]", | |
"id": "#gitc-bwa-mem-samtools.cwl/read_group_header" | |
}, | |
{ | |
"type": { | |
"type": "array", | |
"items": "File" | |
}, | |
"inputBinding": { | |
"position": 3 | |
}, | |
"id": "#gitc-bwa-mem-samtools.cwl/reads" | |
}, | |
{ | |
"type": "File", | |
"inputBinding": { | |
"position": 2 | |
}, | |
"secondaryFiles": [ | |
".amb", | |
".ann", | |
".bwt", | |
".pac", | |
".sa" | |
], | |
"id": "#gitc-bwa-mem-samtools.cwl/reference" | |
}, | |
{ | |
"type": "boolean", | |
"default": false, | |
"inputBinding": { | |
"prefix": "-p" | |
}, | |
"doc": "smart pairing (ignoring in2.fq)", | |
"id": "#gitc-bwa-mem-samtools.cwl/smart_pairing" | |
}, | |
{ | |
"type": "boolean", | |
"default": false, | |
"inputBinding": { | |
"prefix": "-Y" | |
}, | |
"id": "#gitc-bwa-mem-samtools.cwl/soft_clipping" | |
}, | |
{ | |
"type": "int", | |
"inputBinding": { | |
"position": 1, | |
"prefix": "-t" | |
}, | |
"doc": "-t INT number of threads [1]", | |
"id": "#gitc-bwa-mem-samtools.cwl/threads" | |
}, | |
{ | |
"type": [ | |
"null", | |
"int" | |
], | |
"inputBinding": { | |
"prefix": "-v" | |
}, | |
"doc": "verbose level: 1=error, 2=warning, 3=message, 4+=debugging [3]", | |
"id": "#gitc-bwa-mem-samtools.cwl/verbosity" | |
} | |
], | |
"stdout": "$(inputs.output_filename)", | |
"outputs": [ | |
{ | |
"type": "File", | |
"outputBinding": { | |
"glob": "$(inputs.output_filename)" | |
}, | |
"id": "#gitc-bwa-mem-samtools.cwl/output" | |
} | |
], | |
"id": "#gitc-bwa-mem-samtools.cwl" | |
}, | |
{ | |
"class": "CommandLineTool", | |
"requirements": [ | |
{ | |
"class": "DockerRequirement", | |
"dockerPull": "dukegcb/picard:2.10.7" | |
}, | |
{ | |
"class": "InlineJavascriptRequirement" | |
} | |
], | |
"hints": [ | |
{ | |
"class": "SoftwareRequirement", | |
"packages": [ | |
{ | |
"version": [ | |
"2.10.7" | |
], | |
"package": "picard", | |
"https://schema.org/citation": "http://broadinstitute.github.io/picard" | |
} | |
] | |
} | |
], | |
"inputs": [ | |
{ | |
"type": [ | |
"null", | |
{ | |
"type": "array", | |
"items": "File" | |
} | |
], | |
"doc": "The bed file to be converted to interval_list format. Required.", | |
"inputBinding": { | |
"prefix": "I=", | |
"shellQuote": false | |
}, | |
"id": "#picard-BedToIntervalList.cwl/input_file" | |
}, | |
{ | |
"type": [ | |
"null", | |
"string" | |
], | |
"doc": "Interval list output filename.", | |
"default": "list.interval_list", | |
"inputBinding": { | |
"prefix": "O=", | |
"shellQuote": false | |
}, | |
"id": "#picard-BedToIntervalList.cwl/output_filename" | |
}, | |
{ | |
"type": "File", | |
"doc": "The reference sequences in fasta format.", | |
"inputBinding": { | |
"prefix": "SD=", | |
"shellQuote": false | |
}, | |
"id": "#picard-BedToIntervalList.cwl/reference_sequence" | |
} | |
], | |
"outputs": [ | |
{ | |
"type": "File", | |
"outputBinding": { | |
"glob": "$(inputs.output_filename)" | |
}, | |
"id": "#picard-BedToIntervalList.cwl/output_interval_list_file" | |
} | |
], | |
"baseCommand": [ | |
"java", | |
"-Xmx4g" | |
], | |
"arguments": [ | |
{ | |
"valueFrom": "/opt/picard/picard.jar", | |
"position": -1, | |
"prefix": "-jar" | |
}, | |
{ | |
"valueFrom": "BedToIntervalList", | |
"position": 0 | |
} | |
], | |
"id": "#picard-BedToIntervalList.cwl" | |
}, | |
{ | |
"class": "CommandLineTool", | |
"hints": [ | |
{ | |
"class": "DockerRequirement", | |
"dockerPull": "dukegcb/trim-galore:0.4.4" | |
}, | |
{ | |
"class": "SoftwareRequirement", | |
"packages": [ | |
{ | |
"version": [ | |
"1.14" | |
], | |
"package": "cutadapt", | |
"https://schema.org/citation": "https://dx.doi.org/10.14806/ej.17.1.200" | |
}, | |
{ | |
"version": [ | |
"0.4.4" | |
], | |
"package": "trimgalore", | |
"https://schema.org/citation": "https://www.bioinformatics.babraham.ac.uk/projects/trim_galore/" | |
} | |
] | |
} | |
], | |
"inputs": [ | |
{ | |
"type": "boolean", | |
"inputBinding": { | |
"position": 1, | |
"prefix": "--paired" | |
}, | |
"id": "#trim_galore.cwl/paired" | |
}, | |
{ | |
"type": { | |
"type": "array", | |
"items": "File" | |
}, | |
"inputBinding": { | |
"position": 3 | |
}, | |
"id": "#trim_galore.cwl/reads" | |
} | |
], | |
"outputs": [ | |
{ | |
"type": { | |
"type": "array", | |
"items": "File" | |
}, | |
"outputBinding": { | |
"glob": "*_trimming_report.txt" | |
}, | |
"id": "#trim_galore.cwl/trim_reports" | |
}, | |
{ | |
"type": { | |
"type": "array", | |
"items": "File" | |
}, | |
"outputBinding": { | |
"glob": "*_val_*.fq*" | |
}, | |
"id": "#trim_galore.cwl/trimmed_reads" | |
} | |
], | |
"baseCommand": "trim_galore", | |
"arguments": [ | |
{ | |
"valueFrom": "$(runtime.outdir)", | |
"prefix": "-o", | |
"position": 2 | |
} | |
], | |
"id": "#trim_galore.cwl" | |
}, | |
{ | |
"class": "Workflow", | |
"label": "WES GATK4 Preprocessing", | |
"doc": "Whole Exome Sequence analysis GATK4 Preprocessing\n", | |
"requirements": [ | |
{ | |
"class": "ScatterFeatureRequirement" | |
}, | |
{ | |
"class": "SubworkflowFeatureRequirement" | |
}, | |
{ | |
"$import": "#bespin-types.yml" | |
} | |
], | |
"inputs": [ | |
{ | |
"type": [ | |
"null", | |
{ | |
"type": "array", | |
"items": "File" | |
} | |
], | |
"id": "#main/bait_intervals" | |
}, | |
{ | |
"type": [ | |
"null", | |
"int" | |
], | |
"id": "#main/interval_padding" | |
}, | |
{ | |
"type": { | |
"type": "array", | |
"items": "File" | |
}, | |
"secondaryFiles": [ | |
".idx" | |
], | |
"id": "#main/known_sites" | |
}, | |
{ | |
"type": "string", | |
"id": "#main/library" | |
}, | |
{ | |
"type": "string", | |
"id": "#main/platform" | |
}, | |
{ | |
"type": "#bespin-types.yml/FASTQReadPairType", | |
"id": "#main/read_pair" | |
}, | |
{ | |
"type": "File", | |
"secondaryFiles": [ | |
".amb", | |
".ann", | |
".bwt", | |
".pac", | |
".sa", | |
".fai", | |
"^.dict" | |
], | |
"id": "#main/reference_genome" | |
}, | |
{ | |
"type": "File", | |
"secondaryFiles": [ | |
".idx" | |
], | |
"id": "#main/resource_dbsnp" | |
}, | |
{ | |
"type": [ | |
"null", | |
{ | |
"type": "array", | |
"items": "File" | |
} | |
], | |
"id": "#main/target_intervals" | |
}, | |
{ | |
"type": "int", | |
"id": "#main/threads" | |
} | |
], | |
"outputs": [ | |
{ | |
"type": { | |
"type": "array", | |
"items": "File" | |
}, | |
"outputSource": "#main/preprocessing/fastqc_reports", | |
"id": "#main/fastqc_reports" | |
}, | |
{ | |
"type": "File", | |
"outputSource": "#main/preprocessing/haplotypes_bam", | |
"doc": "BAM file containing assembled haplotypes", | |
"id": "#main/haplotypes_bam" | |
}, | |
{ | |
"type": "File", | |
"outputSource": "#main/preprocessing/markduplicates_bam", | |
"doc": "BAM and bai files from markduplicates", | |
"id": "#main/markduplicates_bam" | |
}, | |
{ | |
"type": "File", | |
"outputSource": "#main/preprocessing/raw_variants", | |
"doc": "Variants from HaplotypeCaller", | |
"id": "#main/raw_variants" | |
}, | |
{ | |
"type": "File", | |
"outputSource": "#main/preprocessing/recalibrated_reads", | |
"doc": "BAM file containing recalibrated reads", | |
"id": "#main/recalibrated_reads" | |
}, | |
{ | |
"type": "File", | |
"outputSource": "#main/preprocessing/recalibration_table", | |
"doc": "Table of recalibration", | |
"id": "#main/recalibration_table" | |
}, | |
{ | |
"type": { | |
"type": "array", | |
"items": "File" | |
}, | |
"outputSource": "#main/preprocessing/trim_reports", | |
"id": "#main/trim_reports" | |
} | |
], | |
"steps": [ | |
{ | |
"run": "#exomeseq-00-prepare-reference-data.cwl", | |
"in": [ | |
{ | |
"source": "#main/bait_intervals", | |
"id": "#main/prepare_reference_data/bait_intervals" | |
}, | |
{ | |
"source": "#main/reference_genome", | |
"id": "#main/prepare_reference_data/reference_genome" | |
}, | |
{ | |
"source": "#main/target_intervals", | |
"id": "#main/prepare_reference_data/target_intervals" | |
} | |
], | |
"out": [ | |
"#main/prepare_reference_data/target_interval_list", | |
"#main/prepare_reference_data/bait_interval_list" | |
], | |
"id": "#main/prepare_reference_data" | |
}, | |
{ | |
"run": "#exomeseq-gatk4-01-preprocessing.cwl", | |
"in": [ | |
{ | |
"source": "#main/prepare_reference_data/bait_interval_list", | |
"id": "#main/preprocessing/bait_interval_list" | |
}, | |
{ | |
"source": "#main/interval_padding", | |
"id": "#main/preprocessing/interval_padding" | |
}, | |
{ | |
"source": "#main/target_intervals", | |
"id": "#main/preprocessing/intervals" | |
}, | |
{ | |
"source": "#main/known_sites", | |
"id": "#main/preprocessing/known_sites" | |
}, | |
{ | |
"source": "#main/library", | |
"id": "#main/preprocessing/library" | |
}, | |
{ | |
"source": "#main/platform", | |
"id": "#main/preprocessing/platform" | |
}, | |
{ | |
"source": "#main/read_pair", | |
"id": "#main/preprocessing/read_pair" | |
}, | |
{ | |
"source": "#main/reference_genome", | |
"id": "#main/preprocessing/reference_genome" | |
}, | |
{ | |
"source": "#main/resource_dbsnp", | |
"id": "#main/preprocessing/resource_dbsnp" | |
}, | |
{ | |
"source": "#main/prepare_reference_data/target_interval_list", | |
"id": "#main/preprocessing/target_interval_list" | |
}, | |
{ | |
"source": "#main/threads", | |
"id": "#main/preprocessing/threads" | |
} | |
], | |
"out": [ | |
"#main/preprocessing/fastqc_reports", | |
"#main/preprocessing/trim_reports", | |
"#main/preprocessing/markduplicates_bam", | |
"#main/preprocessing/recalibration_table", | |
"#main/preprocessing/recalibrated_reads", | |
"#main/preprocessing/raw_variants", | |
"#main/preprocessing/haplotypes_bam" | |
], | |
"id": "#main/preprocessing" | |
} | |
], | |
"id": "#main" | |
} | |
], | |
"cwlVersion": "v1.0", | |
"$schemas": [ | |
"https://schema.org/docs/schema_org_rdfa.html" | |
] | |
} |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment