Add Mutect3 dataset generation to Mutect2 WDL

scripts/m2_cromwell_tests/mutect2.inputs.json

@@ -4,10 +4,6 @@
   "Mutect2.intervals": "gs://gatk-best-practices/somatic-b37/whole_exome_agilent_1.1_refseq_plus_3_boosters.Homo_sapiens_assembly19.baits.interval_list",
   "Mutect2.scatter_count": 50,
   "Mutect2.m2_extra_args": "--downsampling-stride 20 --max-reads-per-alignment-start 6 --max-suspicious-reads-per-alignment-start 6",
-  "Mutect2.filter_funcotations": "True",
-  "Mutect2.funco_reference_version": "hg19",
-  "Mutect2.funco_data_sources_tar_gz": "gs://broad-public-datasets/funcotator/funcotator_dataSources.v1.6.20190124s.tar.gz",
-  "Mutect2.funco_transcript_selection_list": "gs://broad-public-datasets/funcotator/transcriptList.exact_uniprot_matches.AKT1_CRLF2_FGFR1.txt",
 
   "Mutect2.ref_fasta": "gs://gatk-best-practices/somatic-b37/Homo_sapiens_assembly19.fasta",
   "Mutect2.ref_dict": "gs://gatk-best-practices/somatic-b37/Homo_sapiens_assembly19.dict",

scripts/m2_cromwell_tests/run_m2_wdl.sh

@@ -12,7 +12,7 @@ echo "Creating tar.gz for Funcotator datasources =========="
 pushd .
 FUNCOTATOR_TEST_DS_DIR=${WORKING_DIR}/gatk/src/test/resources/large/funcotator/
 cd ${FUNCOTATOR_TEST_DS_DIR}
-# First parameter must match Mutect2_Multi.funco_data_sources_tar_gz test_m2_wdl_multi.json
+# First parameter must match Mutect2_Multi.funco_data_sources_tar_gz test_m2_wdl.json
 tar zcvf ${WORKING_DIR}/gatk/small_ds_pik3ca.tar.gz small_ds_pik3ca/*
 popd
 
@@ -35,24 +35,24 @@ fi
 echo "Docker build done =========="
 echo "Putting the newly built docker image into the json parameters"
 cd $WORKING_DIR/gatk/scripts/
-sed -r "s/__GATK_DOCKER__/broadinstitute\/gatk\:$HASH_TO_USE/g" m2_cromwell_tests/test_m2_wdl_multi.json >$WORKING_DIR/test_m2_wdl_multi_mod.json
+sed -r "s/__GATK_DOCKER__/broadinstitute\/gatk\:$HASH_TO_USE/g" m2_cromwell_tests/test_m2_wdl.json >$WORKING_DIR/test_m2_wdl_mod.json
 echo "JSON FILE (modified) ======="
-cat $WORKING_DIR/test_m2_wdl_multi_mod.json
+cat $WORKING_DIR/test_m2_wdl_mod.json
 sed -r "s/__GATK_DOCKER__/broadinstitute\/gatk\:$HASH_TO_USE/g" m2_cromwell_tests/test_mitochondria_m2_wdl.json >$WORKING_DIR/test_mitochondria_m2_wdl_mod.json
 echo "JSON FILE (modified) ======="
 cat $WORKING_DIR/test_mitochondria_m2_wdl_mod.json
 echo "=================="
 
-# Create the tumor-only json by using the pair_list_tumor_only file
-sed -r "s/\"pair_list/\"pair_list_tumor_only/g" $WORKING_DIR/test_m2_wdl_multi_mod.json >$WORKING_DIR/test_m2_wdl_multi_mod_to.json
+# Create the tumor-only json by removing normal_reads and normal_reads_index from the input json
+grep -v 'Mutect2.normal_reads' $WORKING_DIR/test_m2_wdl_mod.json >$WORKING_DIR/test_m2_wdl_mod_to.json
 cd $WORKING_DIR/
 
 echo "Running M2 WDL through cromwell (T/N)"
 ln -fs $WORKING_DIR/gatk/scripts/mutect2_wdl/mutect2.wdl
-sudo java -jar $CROMWELL_JAR run $WORKING_DIR/gatk/scripts/mutect2_wdl/mutect2_multi_sample.wdl -i $WORKING_DIR/test_m2_wdl_multi_mod.json -m $WORKING_DIR/test_m2_wdl.metadata
+sudo java -jar $CROMWELL_JAR run $WORKING_DIR/gatk/scripts/mutect2_wdl/mutect2.wdl -i $WORKING_DIR/test_m2_wdl_mod.json -m $WORKING_DIR/test_m2_wdl.metadata
 
 echo "Running M2 WDL through cromwell (Tumor-only)"
-sudo java -jar $CROMWELL_JAR run $WORKING_DIR/gatk/scripts/mutect2_wdl/mutect2_multi_sample.wdl -i $WORKING_DIR/test_m2_wdl_multi_mod_to.json -m $WORKING_DIR/test_m2_wdl_to.metadata
+sudo java -jar $CROMWELL_JAR run $WORKING_DIR/gatk/scripts/mutect2_wdl/mutect2.wdl -i $WORKING_DIR/test_m2_wdl_mod_to.json -m $WORKING_DIR/test_m2_wdl_to.metadata
 
 echo "Running Mitochondria M2 WDL through cromwell"
 ln -fs $WORKING_DIR/gatk/scripts/mitochondria_m2_wdl/AlignAndCall.wdl

scripts/m2_cromwell_tests/test_m2_wdl.json

@@ -0,0 +1,16 @@
+{
+  "Mutect2.gatk_docker": "__GATK_DOCKER__",
+  "Mutect2.intervals": "/home/runner/work/gatk/gatk/scripts/m2_cromwell_tests/interval_list.interval_list",
+  "Mutect2.ref_fasta": "/home/runner/work/gatk/gatk/src/test/resources/large/human_g1k_v37.20.21.fasta",
+  "Mutect2.ref_fai": "/home/runner/work/gatk/gatk/src/test/resources/large/human_g1k_v37.20.21.fasta.fai",
+  "Mutect2.ref_dict": "/home/runner/work/gatk/gatk/src/test/resources/large/human_g1k_v37.20.21.dict",
+  "Mutect2.tumor_reads": "/home/runner/work/gatk/gatk/src/test/resources/large/mutect/dream_synthetic_bams/tumor_1.bam",
+  "Mutect2.tumor_reads_index": "/home/runner/work/gatk/gatk/src/test/resources/large/mutect/dream_synthetic_bams/tumor_1.bam.bai",
+  "Mutect2.normal_reads": "/home/runner/work/gatk/gatk/src/test/resources/large/mutect/dream_synthetic_bams/normal_1.bam",
+  "Mutect2.normal_reads_index": "/home/runner/work/gatk/gatk/src/test/resources/large/mutect/dream_synthetic_bams/normal_1.bam.bai",
+  "Mutect2.scatter_count": 2,
+  "Mutect2.run_orientation_bias_mixture_model_filter": true,
+  "Mutect2.preemptible": 2,
+  "Mutect2.compress_vcfs": false,
+  "Mutect2.make_bamout": true
+}