Create full test without GATK GenomicsDBImport which is not supported…

… on ARM
wtsi-ssg · Aug 22, 2023 · aeceeaa · aeceeaa
1 parent 934e0d2
commit aeceeaa
Showing 1 changed file with 141 additions and 0 deletions.
diff --git a/setup/config_files/full_test_arm.yml b/setup/config_files/full_test_arm.yml
@@ -0,0 +1,141 @@
+---
+- general_settings:
+  install_dir: "/data/tools/"
+- type: cpu
+  benchmarks:
+    - type: multithreaded 
+      settings:
+        tag: bwa
+        command: "${path_to_program} mem -t ${threads} -K 100000000 -o ${result_path}/output.${repeatn}.${processn}.sam ${input_datapath}/GCA_000001405.15_GRCh38_full_plus_hs38d1_analysis_set.fna ${input_datapath}/RMNISTHS_3xdownsample.fastq"
+        program: "bwa"
+        programversion: "0.7.17+139f68f"
+        shell: False
+        dataset_file: "/benchmarking/setup/bwa_data.txt"
+        datadir: "/data/datasets/bwa/"
+        result_dir: "/data/results/runs/"
+        repeats: 3
+        units: 3x WGS
+        process_thread:
+          - processes: 2
+            threads: $N/2
+          - processes: 4
+            threads: $N/4
+          - processes: 1
+            threads: range(2,$N,4)
+        clear_caches: True
+    - type: multithreaded 
+      settings:
+        tag: bwa-mem2
+        command: "${path_to_program} mem -t ${threads} -K 100000000 -o ${result_path}/output.${repeatn}.${processn}.sam ${input_datapath}/GCA_000001405.15_GRCh38_full_plus_hs38d1_analysis_set.fna ${input_datapath}/RMNISTHS_3xdownsample.fastq"
+        program: "bwa-mem2"
+        programversion: "2.2.1+simde_benchmarking"
+        shell: False
+        dataset_file: "/benchmarking/setup/bwa_data.txt"
+        datadir: "/data/datasets/bwa/"
+        result_dir: "/data/results/runs/"
+        repeats: 3
+        units: 3x WGS
+        process_thread:
+          - processes: 2
+            threads: $N/2
+          - processes: 4
+            threads: $N/4
+          - processes: 1
+            threads: range(2,$N,4)
+        clear_caches: True
+    - type: multithreaded
+      settings:
+        tag: salmon
+        command: "${path_to_program} quant --threads ${threads} -i ${input_datapath}/referenceGenome_index4/ -l A -1 ${input_datapath}/SRR10103759/C1_1.fq.gz -2 ${input_datapath}/SRR10103759/C1_2.fq.gz -o ${result_path}output.${repeatn}.${processn}.quant.sf"
+        program: "salmon"
+        programversion: "1.10.1"
+        shell: False
+        dataset_file:
+          file: "/benchmarking/setup/salmon_input_data.txt"
+          dest: "SRR10103759"
+        dataset_tag: "default"
+        datadir: "/data/datasets/salmon/"
+        result_dir: "/data/results/runs/"
+        repeats: 3
+        units: run
+        process_thread:
+          - processes: 1
+            threads: $N
+          - processes: 2
+            threads: $N/2
+          - processes: 4
+            threads: $N/4
+        clear_caches: True
+    - type: mbw
+      settings:
+        program: "mbw"
+        programversion: "1.4a"
+        arguments: "512MiB"
+    - type: multithreaded
+      settings:
+        tag: bcftools-1.15.1
+        command: "${path_to_program} mpileup --threads ${threads} -Ou -f ${input_datapath}/b37_human_decoy_reference.fasta ${input_datapath}/RMNISTHS_30xdownsample_chr22.cram | ${install_path}bcftools call --threads ${threads} -mv -o ${result_path}output.${repeatn}.${processn}.vcf.gz"
+        program: "bcftools"
+        programversion: "1.15.1"
+        shell: True
+        dataset_file: "/benchmarking/setup/variantcaller_data.txt"
+        datadir: "/data/datasets/variantcaller/"
+        result_dir: "/data/results/runs/"
+        repeats: 3
+        units: Chr22 @ 30x from WGS
+        process_thread:
+          - processes: 1
+            threads: $N
+          - processes: 2
+            threads: $N/2
+          - processes: 4
+            threads: $N/4
+        clear_caches: True
+    - type: multithreaded
+      settings:
+        tag: minimap2
+        command: "${path_to_program} -x ava-pb -t ${threads} ${input_datapath}/pacbio_filtered.fastq ${input_datapath}/pacbio_filtered.fastq | gzip -1 > ${result_path}output.${repeatn}.${processn}.paf.gz"
+        program: "minimap2"
+        programversion: "2.26+24d0af6"
+        shell: True
+        dataset_file: "/benchmarking/setup/minimapasm_data.txt"
+        datadir: "/data/datasets/minimapasm/"
+        result_dir: "/data/results/runs/"
+        repeats: 3
+        units: random subset of 30X E.coli from a PacBio RS II P5-C3 single SMRTcell
+        process_thread:
+          - processes: 2
+            threads: $N/2
+          - processes: 1
+            threads: $N
+          - processes: 4
+            threads: $N/4
+        clear_caches: True
+    - type: multithreaded
+      settings:
+        tag: miniasm
+        command: "${path_to_program} -f ${input_datapath}/pacbio_filtered.fastq ${input_datapath}/pacbio_filtered.paf.gz > ${result_path}output.${repeatn}.${processn}.gfa"
+        program: "miniasm"
+        programversion: "0.3+ce615d1"
+        shell: True
+        dataset_file: "/benchmarking/setup/minimapasm_data.txt"
+        datadir: "/data/datasets/minimapasm/"
+        result_dir: "/data/results/runs/"
+        repeats: 3
+        units: random subset of 30X E.coli from a PacBio RS II P5-C3 single SMRTcell
+        process_thread:
+          - processes: 2
+            threads: 1
+          - processes: 1
+            threads: 1
+          - processes: 4
+            threads: 1
+        clear_caches: True
+- type: disk
+  benchmarks:
+    - type: iozone
+      settings:
+        program: "iozone"
+        programversion: "3.492"
+        arguments: "-a"
+        result_dir: "/data/results/runs"