Test cwl results (#94)

* add test for CollectQualityYieldMetrics

* add test for CollectReadgroupBamQualityMetrics

* add test for deepvariant

Makefile

@@ -141,6 +141,9 @@ test: $(PYSOURCES) cwltest
 cwltest: wdl2cwl/tests/cwl_files/*.cwl wdl2cwl/tests/*
 	cd wdl2cwl/tests && cwltest --tool cwltool --test cwl_tests.yaml
 
+cwltest_big: wdl2cwl/tests/cwl_files/*.cwl wdl2cwl/tests/*
+	cd wdl2cwl/tests && cwltest --tool cwltool --test cwl_tests_big.yaml
+
 ## testcov     : run the ${MODULE} test suite and collect coverage
 testcov: $(PYSOURCES)
 	pytest --cov

wdl2cwl/tests/cwl_files/CollectQualityYieldMetrics.cwl

@@ -25,6 +25,7 @@ requirements:
               INPUT=$(inputs.input_bam.path) \
               OQ=true \
               OUTPUT=$(inputs.metrics_filename)
+            sed -i -e 1,5d $(inputs.metrics_filename)  # for reproducibility
   - class: InlineJavascriptRequirement
   - class: NetworkAccess
     networkAccess: true

wdl2cwl/tests/cwl_files/CollectReadgroupBamQualityMetrics.cwl

@@ -59,6 +59,7 @@ requirements:
               $(inputs.collect_gc_bias_metrics ? 'PROGRAM="CollectGcBiasMetrics"' : "") \
               METRIC_ACCUMULATION_LEVEL=null \
               METRIC_ACCUMULATION_LEVEL=READ_GROUP
+            sed -i -e 1,5d "$(inputs.output_bam_prefix).alignment_summary_metrics"   # for reproducibility
   - class: InlineJavascriptRequirement
   - class: NetworkAccess
     networkAccess: true

wdl2cwl/tests/cwl_files/deepvariant.cwl

@@ -34,9 +34,7 @@ inputs:
       - string
       - 'null'
   - id: regions
-    type:
-      - File
-      - 'null'
+    type: string
   - id: sampleName
     type:
       - string
@@ -91,15 +89,21 @@ requirements:
         entry: |4
 
             set -e
+            mkdir reference_dir
+            ln -s $(inputs.referenceFasta.path) reference_dir/\$(basename $(inputs.referenceFasta.path))
+            ln -s $(inputs.referenceFastaIndex.path) reference_dir/\$(basename $(inputs.referenceFastaIndex.path))
+            mkdir bam_dir
+            ln -s $(inputs.inputBam.path) bam_dir/\$(basename $(inputs.inputBam.path))
+            ln -s $(inputs.inputBamIndex.path) bam_dir/\$(basename $(inputs.inputBamIndex.path))
             /opt/deepvariant/bin/run_deepvariant \
-            --ref $(inputs.referenceFasta.path) \
-            --reads $(inputs.inputBam.path) \
+            --ref reference_dir/\$(basename $(inputs.referenceFasta.path)) \
+            --reads bam_dir/\$(basename $(inputs.inputBam.path)) \
             --model_type $(inputs.modelType) \
             --output_vcf $(inputs.outputVcf) \
             $(inputs.outputGVcf === null ? "" : "--output_gvcf " + inputs.outputGVcf ) \
             $(inputs.customizedModel === null ? "" : "--customized_model " + inputs.customizedModel.path ) \
             $(inputs.numShards === null ? "" : "--num_shards " + inputs.numShards ) \
-            $(inputs.regions === null ? "" : "--regions " + inputs.regions.path ) \
+            --regions $(inputs.regions) \
             $(inputs.sampleName === null ? "" : "--sample_name " + inputs.sampleName ) \
             $(inputs.postprocessVariantsExtraArgs === null ? "" : "--postprocess_variants_extra_args " + inputs.postprocessVariantsExtraArgs ) \
             $(inputs.VCFStatsReport ? "--vcf_stats_report" : "--novcf_stats_report")

wdl2cwl/tests/cwl_tests.yaml

@@ -5,6 +5,15 @@
   job: inputs/validateOptimus_2.yaml
   output:
     result: PASS
+- doc: CollectQualityYieldMetrics
+  tool: cwl_files/CollectQualityYieldMetrics.cwl
+  job: inputs/CollectQualityYieldMatrics.yaml
+  output:
+    quality_yield_metrics:
+      basename: collect_quality_yield_metrics.quals.txt
+      checksum: sha1$870d2ec27fb38292079c8b135fdf864c03f32bfe
+      class: File
+      size: 261
 
 # Tests for doc: bcftools_stats
 - doc: bcftools_stats

wdl2cwl/tests/inputs/CollectQualityYieldMatrics.yaml

@@ -0,0 +1,9 @@
+input_bam:
+  class: File
+  path: valid.bam
+  doc: |
+    From https://github.com/broadinstitute/gatk/blob/7f96bf6677040851c6da3cb96d97f957f1df05bc/src/test/resources/org/broadinstitute/hellbender/metrics/analysis/CollectQualityYieldMetrics/valid.bam
+    Copyright 2021 Broad Institute, Inc.
+    Licensed under the Apache License, Version 2.0 (the "License")
+metrics_filename: collect_quality_yield_metrics.quals.txt
+preemptible_tries: 0

wdl2cwl/tests/inputs/CollectReadgroupBamQualityMetrics.yaml

@@ -0,0 +1,22 @@
+input_bam:
+  class: File
+  path: RP-929.NA12878.bam
+  doc: created from running https://github.com/mr-c/warp/blob/more_public_data/pipelines/broad/reprocessing/external/exome/test_inputs/Plumbing/RP-929.NA12878.json
+input_bam_index:
+  class: File
+  path: RP-929.NA12878.bai 
+  doc: created from running https://github.com/mr-c/warp/blob/more_public_data/pipelines/broad/reprocessing/external/exome/test_inputs/Plumbing/RP-929.NA12878.json
+output_bam_prefix: RP-929.NA12878.readgroup
+ref_dict:
+  class: File
+  location: https://storage.googleapis.com/gcp-public-data--broad-references/hg38/v0/Homo_sapiens_assembly38.dict
+ref_fasta:
+  class: File
+  path: Homo_sapiens_assembly38.fasta
+  location: https://storage.googleapis.com/gcp-public-data--broad-references/hg38/v0/Homo_sapiens_assembly38.fasta
+ref_fasta_index:
+  class: File
+  path: Homo_sapiens_assembly38.fasta
+  location: https://storage.googleapis.com/gcp-public-data--broad-references/hg38/v0/Homo_sapiens_assembly38.fasta.fai
+collect_gc_bias_metrics: False
+preemptible_tries: 3

wdl2cwl/tests/inputs/deepvariant-testdata/test_nist.b37_chr20_100kbp_at_10mb.bed

@@ -0,0 +1,11 @@
+chr20	10000846	10002407
+chr20	10002520	10004171
+chr20	10004273	10004964
+chr20	10004994	10006386
+chr20	10006409	10007800
+chr20	10007824	10008018
+chr20	10008043	10008079
+chr20	10008100	10008707
+chr20	10008808	10008897
+chr20	10009002	10009791
+chr20	10009933	10010531