From ec300c76b5a3f927c3cd5035669b70847ade3854 Mon Sep 17 00:00:00 2001 From: Sarah Wait Zaranek Date: Thu, 23 Jul 2020 13:09:38 +0000 Subject: [PATCH] Updating cwl and yml for intervals and other changes Arvados-DCO-1.1-Signed-off-by: Sarah Wait Zaranek no issue # --- .../cwl/helper/bwamem-gatk-report-wf.cwl | 4 ++++ .../cwl/helper/gatk-splitintervals.cwl | 5 +++++ WGS-processing/cwl/helper/getfastq.cwl | 1 + .../cwl/helper/scatter-gatk-wf-with-interval.cwl | 4 ++++ WGS-processing/cwl/wgs-processing-wf.cwl | 3 +++ .../yml/helper/bwamem-gatk-report-wf.yml | 4 ++++ .../yml/helper/bwamem-samtools-view-test.yml | 16 ++++++++++++++++ .../yml/helper/gatk-splitintervals.yml | 5 +++++ .../yml/helper/scatter-gatk-wf-with-interval.yml | 4 ++++ WGS-processing/yml/wgs-processing-wf.yml | 8 ++++++-- 10 files changed, 52 insertions(+), 2 deletions(-) create mode 100644 WGS-processing/yml/helper/bwamem-samtools-view-test.yml diff --git a/WGS-processing/cwl/helper/bwamem-gatk-report-wf.cwl b/WGS-processing/cwl/helper/bwamem-gatk-report-wf.cwl index 522bd82..dbb1d50 100644 --- a/WGS-processing/cwl/helper/bwamem-gatk-report-wf.cwl +++ b/WGS-processing/cwl/helper/bwamem-gatk-report-wf.cwl @@ -26,6 +26,9 @@ inputs: - .sa - .fai - ^.dict + fullintervallist: + type: File + label: Full list of intervals to operate over sample: type: string label: Sample Name @@ -110,6 +113,7 @@ steps: run: scatter-gatk-wf-with-interval.cwl in: reference: reference + fullintervallist: fullintervallist bam: samtools-index/indexedbam sample: sample scattercount: scattercount diff --git a/WGS-processing/cwl/helper/gatk-splitintervals.cwl b/WGS-processing/cwl/helper/gatk-splitintervals.cwl index 98758d1..4909bc7 100644 --- a/WGS-processing/cwl/helper/gatk-splitintervals.cwl +++ b/WGS-processing/cwl/helper/gatk-splitintervals.cwl @@ -32,6 +32,9 @@ inputs: - .sa - .fai - ^.dict + fullintervallist: + type: File + label: Full list of intervals to operate over sample: type: string label: Sample Name @@ -54,6 +57,8 @@ arguments: - SplitIntervals - prefix: "-R" valueFrom: $(inputs.reference) + - prefix: "-L" + valueFrom: $(inputs.fullintervallist) - prefix: "--scatter-count" valueFrom: $(inputs.scattercount) - prefix: "--subdivision-mode" diff --git a/WGS-processing/cwl/helper/getfastq.cwl b/WGS-processing/cwl/helper/getfastq.cwl index 035b4c3..59c6771 100644 --- a/WGS-processing/cwl/helper/getfastq.cwl +++ b/WGS-processing/cwl/helper/getfastq.cwl @@ -41,6 +41,7 @@ expression: | var fastq2 = []; for (var i = 0; i < inputs.fastqdir.listing.length; i++) { var name = inputs.fastqdir.listing[i]; + name.format = 'http://edamontology.org/format_1930' if (name.basename.indexOf('_1.fastq.gz') != -1 ) { fastq1.push(name); } diff --git a/WGS-processing/cwl/helper/scatter-gatk-wf-with-interval.cwl b/WGS-processing/cwl/helper/scatter-gatk-wf-with-interval.cwl index 9752edb..80d7601 100644 --- a/WGS-processing/cwl/helper/scatter-gatk-wf-with-interval.cwl +++ b/WGS-processing/cwl/helper/scatter-gatk-wf-with-interval.cwl @@ -25,6 +25,9 @@ inputs: - .sa - .fai - ^.dict + fullintervallist: + type: File + label: Full list of intervals to operate over sample: type: string label: Sample Name @@ -52,6 +55,7 @@ steps: run: gatk-splitintervals.cwl in: reference: reference + fullintervallist: fullintervallist sample: sample scattercount: scattercount out: [intervalfiles] diff --git a/WGS-processing/cwl/wgs-processing-wf.cwl b/WGS-processing/cwl/wgs-processing-wf.cwl index 7a5e6de..8cbcf5e 100644 --- a/WGS-processing/cwl/wgs-processing-wf.cwl +++ b/WGS-processing/cwl/wgs-processing-wf.cwl @@ -22,6 +22,8 @@ inputs: - .sa - .fai - ^.dict + fullintervallist: + type: File knownsites: type: File format: edam:format_3016 # VCF @@ -74,6 +76,7 @@ steps: fastq1: getfastq/fastq1 fastq2: getfastq/fastq2 reference: reference + fullintervallist: fullintervallist sample: getfastq/sample knownsites: knownsites scattercount: scattercount diff --git a/WGS-processing/yml/helper/bwamem-gatk-report-wf.yml b/WGS-processing/yml/helper/bwamem-gatk-report-wf.yml index 5e5be2c..8a92726 100644 --- a/WGS-processing/yml/helper/bwamem-gatk-report-wf.yml +++ b/WGS-processing/yml/helper/bwamem-gatk-report-wf.yml @@ -5,6 +5,10 @@ reference: format: edam:format_1929 location: keep:a3af04432df3d71d22f2fe8be549ba96+5974/hg38.fa +fullintervallist: + class: File + location: /data-sdd/pgp/testdata/full-noalt-scattered.interval_list + fastq1: class: File format: edam:format_1930 diff --git a/WGS-processing/yml/helper/bwamem-samtools-view-test.yml b/WGS-processing/yml/helper/bwamem-samtools-view-test.yml new file mode 100644 index 0000000..55b4787 --- /dev/null +++ b/WGS-processing/yml/helper/bwamem-samtools-view-test.yml @@ -0,0 +1,16 @@ +reference: + class: File + format: edam:format_1929 + location: keep:5eac1d1c03ff6404226421a8d8351cf1+6133/GRCh38_no_alt_plus_hs38d1_analysis_set.fna + +fastq1: + class: File + format: edam:format_1930 + location: keep:24df1ea93ad5275aa511462ea85b0548+20283/WGC071838D_R1.fastq.gz + +fastq2: + class: File + format: edam:format_1930 + location: keep:24df1ea93ad5275aa511462ea85b0548+20283/WGC071838D_R2.fastq.gz + +sample: WGC071838D diff --git a/WGS-processing/yml/helper/gatk-splitintervals.yml b/WGS-processing/yml/helper/gatk-splitintervals.yml index 6c1c24a..e78265c 100644 --- a/WGS-processing/yml/helper/gatk-splitintervals.yml +++ b/WGS-processing/yml/helper/gatk-splitintervals.yml @@ -3,6 +3,11 @@ reference: format: edam:format_1929 location: keep:a3af04432df3d71d22f2fe8be549ba96+5974/hg38.fa +fullintervallist: + class: File + location: /data-sdd/pgp/testdata/full-noalt-scattered.interval_list + sample: ERR1726424 scattercount: "32" + diff --git a/WGS-processing/yml/helper/scatter-gatk-wf-with-interval.yml b/WGS-processing/yml/helper/scatter-gatk-wf-with-interval.yml index a6d7414..20b5a3b 100644 --- a/WGS-processing/yml/helper/scatter-gatk-wf-with-interval.yml +++ b/WGS-processing/yml/helper/scatter-gatk-wf-with-interval.yml @@ -8,6 +8,10 @@ reference: format: edam:format_1929 location: keep:a3af04432df3d71d22f2fe8be549ba96+5974/hg38.fa +fullintervallist: + class: File + location: /data-sdd/pgp/testdata/full-noalt-scattered.interval_list + sample: ERR1726424 scattercount: '32' diff --git a/WGS-processing/yml/wgs-processing-wf.yml b/WGS-processing/yml/wgs-processing-wf.yml index f56f8cb..299a674 100644 --- a/WGS-processing/yml/wgs-processing-wf.yml +++ b/WGS-processing/yml/wgs-processing-wf.yml @@ -1,7 +1,11 @@ reference: class: File format: edam:format_1929 - location: keep:a3af04432df3d71d22f2fe8be549ba96+5974/hg38.fa + location: keep:5eac1d1c03ff6404226421a8d8351cf1+6133/GRCh38_no_alt_plus_hs38d1_analysis_set.fna + +fullintervallist: + class: File + location: /data-sdd/pgp/testdata/full-noalt-scattered.interval_list fastqdir: class: Directory @@ -26,7 +30,7 @@ tailhtml: format: edam:format_2331 location: ../src/annotation/tail.html -scattercount: '32' +scattercount: '20' knownsites: class: File -- 2.30.2