Update defaults (#171)

* update defaults * expose memory * Revert "expose memory" This reverts commit 36b9cda. * update md5sums * adjust md5sums and test inputs * completely empty for test * -s for gambit_db_genomes input so that automated tests pass with 0 byte fake file * update md5sums * update test to not use updated gambit database for cromwell --------- Co-authored-by: kapsakcj <[email protected]>
theiagen · Aug 28, 2023 · 683c84c · 683c84c
1 parent 38770c2
commit 683c84c
Show file tree

Hide file tree

Showing 20 changed files with 50 additions and 41 deletions.
diff --git a/tasks/species_typing/task_pangolin.wdl b/tasks/species_typing/task_pangolin.wdl
@@ -6,7 +6,7 @@ task pangolin4 {
     String samplename
     Int min_length = 10000
     Float max_ambig = 0.5
-    String docker = "us-docker.pkg.dev/general-theiagen/staphb/pangolin:4.3-pdata-1.21"
+    String docker = "us-docker.pkg.dev/general-theiagen/staphb/pangolin:4.3.1-pdata-1.22"
     String? analysis_mode
     Boolean expanded_lineage=true
     Boolean skip_scorpio=false

diff --git a/tasks/task_versioning.wdl b/tasks/task_versioning.wdl
@@ -8,7 +8,7 @@ task version_capture {
     volatile: true
   }
   command {
-    PHB_Version="PHB v1.0.1"
+    PHB_Version="PHB v1.1.0"
     ~{default='' 'export TZ=' + timezone}
     date +"%Y-%m-%d" > TODAY
     echo "$PHB_Version" > PHB_VERSION

diff --git a/tasks/taxon_id/task_gambit.wdl b/tasks/taxon_id/task_gambit.wdl
@@ -5,8 +5,8 @@ task gambit {
     File assembly
     String samplename
     String docker = "us-docker.pkg.dev/general-theiagen/staphb/gambit:0.5.0"
-    File? gambit_db_genomes
-    File? gambit_db_signatures
+    File gambit_db_genomes = "gs://theiagen-public-files/terra/gambit_files/1.1.0/gambit-metadata-1.1-230417.gdb"
+    File gambit_db_signatures = "gs://theiagen-public-files/terra/gambit_files/1.1.0/gambit-signatures-1.1-230417.gs"
     Int disk_size = 100
     Int memory = 16 # set default
     Int cpu = 8 # set default
@@ -20,7 +20,8 @@ task gambit {
     gambit --version | tee GAMBIT_VERSION
 
     # set gambit reference dir; will assume that gambit genomes and signatures will be provided by user in tandem or not at all
-    if [[ ! -z "~{gambit_db_genomes}" ]]; then 
+    # -s evaluates to TRUE if the file exists and has a size greater than zero
+    if [[ -s "~{gambit_db_genomes}" ]]; then 
       echo "User gabmit db identified; ~{gambit_db_genomes} will be utilized for alignment"
       gambit_db_version="$(basename -- '~{gambit_db_genomes}'); $(basename -- '~{gambit_db_signatures}')"
       gambit_db_dir="${PWD}/gambit_database"

diff --git a/tests/inputs/completely-empty-for-test.txt b/tests/inputs/completely-empty-for-test.txt
diff --git a/tests/inputs/theiaprok/wf_theiaprok_illumina_pe.json b/tests/inputs/theiaprok/wf_theiaprok_illumina_pe.json
@@ -34,5 +34,7 @@
     "theiaprok_illumina_pe.merlin_magic.cladetyper.ref_clade4_annotated" : "./tests/inputs/empty-for-test.txt",
     "theiaprok_illumina_pe.merlin_magic.cladetyper.ref_clade5" : "./tests/inputs/empty-for-test.txt",
     "theiaprok_illumina_pe.merlin_magic.cladetyper.ref_clade5_annotated" : "./tests/inputs/empty-for-test.txt",
-    "theiaprok_illumina_pe.bakta.bakta_db" : "./tests/inputs/empty-for-test.txt"
+    "theiaprok_illumina_pe.bakta.bakta_db" : "./tests/inputs/empty-for-test.txt",
+    "theiaprok_illumina_pe.gambit.gambit_db_signatures" : "./tests/inputs/completely-empty-for-test.txt",
+    "theiaprok_illumina_pe.gambit.gambit_db_genomes" : "./tests/inputs/completely-empty-for-test.txt"
 }
diff --git a/tests/inputs/theiaprok/wf_theiaprok_illumina_pe_cromwell.json b/tests/inputs/theiaprok/wf_theiaprok_illumina_pe_cromwell.json
@@ -6,5 +6,7 @@
    "theiaprok_illumina_pe.read_QC_trim.call_midas": false,
   "theiaprok_illumina_pe.genome_annotation": "prokka",
   "theiaprok_illumina_pe.shovill_pe.assembler": "skesa",
-  "theiaprok_illumina_pe.merlin_magic.call_poppunk": false
+  "theiaprok_illumina_pe.merlin_magic.call_poppunk": false,
+  "theiaprok_illumina_pe.gambit.gambit_db_signatures" : "./tests/inputs/completely-empty-for-test.txt",
+  "theiaprok_illumina_pe.gambit.gambit_db_genomes" : "./tests/inputs/completely-empty-for-test.txt"
 }
diff --git a/tests/inputs/theiaprok/wf_theiaprok_illumina_se.json b/tests/inputs/theiaprok/wf_theiaprok_illumina_se.json
@@ -33,5 +33,7 @@
     "theiaprok_illumina_se.merlin_magic.cladetyper.ref_clade4_annotated" : "./tests/inputs/empty-for-test.txt",
     "theiaprok_illumina_se.merlin_magic.cladetyper.ref_clade5" : "./tests/inputs/empty-for-test.txt",
     "theiaprok_illumina_se.merlin_magic.cladetyper.ref_clade5_annotated" : "./tests/inputs/empty-for-test.txt",
-    "theiaprok_illumina_se.bakta.bakta_db" : "./tests/inputs/empty-for-test.txt"
+    "theiaprok_illumina_se.bakta.bakta_db" : "./tests/inputs/empty-for-test.txt",
+    "theiaprok_illumina_se.gambit.gambit_db_signatures" : "./tests/inputs/completely-empty-for-test.txt",
+    "theiaprok_illumina_se.gambit.gambit_db_genomes" : "./tests/inputs/completely-empty-for-test.txt"
 }
diff --git a/tests/inputs/theiaprok/wf_theiaprok_illumina_se_cromwell.json b/tests/inputs/theiaprok/wf_theiaprok_illumina_se_cromwell.json
@@ -5,5 +5,7 @@
   "theiaprok_illumina_se.read_QC_trim.call_midas": false,
   "theiaprok_illumina_se.genome_annotation": "prokka",
   "theiaprok_illumina_se.shovill_se.assembler": "skesa",
-  "theiaprok_illumina_se.merlin_magic.call_poppunk": false
+  "theiaprok_illumina_se.merlin_magic.call_poppunk": false,
+  "theiaprok_illumina_se.gambit.gambit_db_signatures" : "./tests/inputs/completely-empty-for-test.txt",
+  "theiaprok_illumina_se.gambit.gambit_db_genomes" : "./tests/inputs/completely-empty-for-test.txt"
 }
diff --git a/tests/workflows/theiacov/test_wf_theiacov_clearlabs.yml b/tests/workflows/theiacov/test_wf_theiacov_clearlabs.yml
@@ -250,7 +250,7 @@
     - path: miniwdl_run/call-nextclade/work/nextclade_dataset_dir/tag.json
     - path: miniwdl_run/call-nextclade/work/nextclade_dataset_dir/tree.json
     - path: miniwdl_run/call-nextclade/work/nextclade_dataset_dir/virus_properties.json
-      md5sum: 1567bbba6ac947a96db159df7fd102dd
+      md5sum: c0aa53e26cea703a0e7dc8d292ff5c89
     - path: miniwdl_run/call-nextclade/work/nextclade_gene_E.translation.fasta
       md5sum: 14808ad8b34c8bac7de500707400250e
     - path: miniwdl_run/call-nextclade/work/nextclade_gene_M.translation.fasta
@@ -318,15 +318,15 @@
     - path: miniwdl_run/call-pangolin4/work/PANGOLIN_LINEAGE
       md5sum: 717da6cd0df2d2f1d00461f3498aaca9
     - path: miniwdl_run/call-pangolin4/work/PANGOLIN_NOTES
-      md5sum: 3cf7fd185fe58af5952d52e91d4805c0
+      md5sum: 59478efddde2191ead1b46b1f121bbc9
     - path: miniwdl_run/call-pangolin4/work/PANGO_ASSIGNMENT_VERSION
-      md5sum: 4f37bb038d973d7e32df5cbbc4067c87
+      md5sum: edfb0c7ebdbf598df19861b650ee904a
     - path: miniwdl_run/call-pangolin4/work/VERSION_PANGOLIN_ALL
-      md5sum: 6c7f688ada8c1e247782758d7fca08e7
+      md5sum: ef9c730f9b45d2fc284575c13fe7c920
     - path: miniwdl_run/call-pangolin4/work/_miniwdl_inputs/0/clearlabs.medaka.consensus.fasta
       md5sum: d41d8cd98f00b204e9800998ecf8427e
     - path: miniwdl_run/call-pangolin4/work/clearlabs.pangolin_report.csv
-      md5sum: e0512d549aa89d290f6aee20b63cbb63
+      md5sum: 58773dc8fca41e521ec7ab24df371365
     - path: miniwdl_run/call-sc2_gene_coverage/command
       md5sum: e44c070a15a0f8d42348103b0440f691
     - path: miniwdl_run/call-sc2_gene_coverage/inputs.json

diff --git a/tests/workflows/theiacov/test_wf_theiacov_fasta.yml b/tests/workflows/theiacov/test_wf_theiacov_fasta.yml
@@ -38,7 +38,7 @@
       md5sum: 6808ca805661622ad65ae014a4b2a094
     - path: miniwdl_run/call-consensus_qc/work/_miniwdl_inputs/0/clearlabs.fasta.gz
     - path: miniwdl_run/call-nextclade/command
-      md5sum: 847ccfc48474456a512081b2782383dd
+      md5sum: b5ecaad831316b3bd8f066f1e71cc0a5
     - path: miniwdl_run/call-nextclade/inputs.json
     - path: miniwdl_run/call-nextclade/outputs.json
     - path: miniwdl_run/call-nextclade/stderr.txt
@@ -71,9 +71,9 @@
     - path: miniwdl_run/call-nextclade/work/nextclade_dataset_dir/sequences.fasta
       md5sum: bb6b4e9e91304a396724bcb6344b8a5d
     - path: miniwdl_run/call-nextclade/work/nextclade_dataset_dir/tag.json
-      md5sum: 190efe28d393f17d333b3b0e147adb4f
+      md5sum: 97e1309e683fbaaa839198d88cd4e2d9
     - path: miniwdl_run/call-nextclade/work/nextclade_dataset_dir/tree.json
-      md5sum: 78fa5287136e13735b6c903b6a15f0ca
+      md5sum: 6892e6019bf88ec571b4560d66d3acb0
     - path: miniwdl_run/call-nextclade/work/nextclade_dataset_dir/virus_properties.json
     - path: miniwdl_run/call-nextclade/work/nextclade_gene_E.translation.fasta
       md5sum: dc43b1e98245a25c142aec52b29a07df
@@ -139,12 +139,12 @@
     - path: miniwdl_run/call-pangolin4/work/PANGOLIN_NOTES
       md5sum: 71eba5c871bca955ab2a69dbd2c3c62e
     - path: miniwdl_run/call-pangolin4/work/PANGO_ASSIGNMENT_VERSION
-      md5sum: 138848a7965e23fa45633d26abbd2aaf
+      md5sum: 8eb91404219a5758741f83f82e9e328d
     - path: miniwdl_run/call-pangolin4/work/VERSION_PANGOLIN_ALL
-      md5sum: 6c7f688ada8c1e247782758d7fca08e7
+      md5sum: ef9c730f9b45d2fc284575c13fe7c920
     - path: miniwdl_run/call-pangolin4/work/_miniwdl_inputs/0/clearlabs.fasta.gz
     - path: miniwdl_run/call-pangolin4/work/fasta.pangolin_report.csv
-      md5sum: a46fd09018b34d6309126b812ddeb8cd
+      md5sum: e02c65e22296686804e2be23b6d4a696
     - path: miniwdl_run/call-vadr/command
       md5sum: f4ad614b7ad39f28a8145cec280a93c0
     - path: miniwdl_run/call-vadr/inputs.json

diff --git a/tests/workflows/theiacov/test_wf_theiacov_illumina_pe.yml b/tests/workflows/theiacov/test_wf_theiacov_illumina_pe.yml
@@ -368,9 +368,9 @@
     - path: miniwdl_run/call-pangolin4/work/PANGOLIN_NOTES
       md5sum: e98d2fc28664c0622f6b490433286e32
     - path: miniwdl_run/call-pangolin4/work/PANGO_ASSIGNMENT_VERSION
-      md5sum: 4f37bb038d973d7e32df5cbbc4067c87
+      md5sum: edfb0c7ebdbf598df19861b650ee904a
     - path: miniwdl_run/call-pangolin4/work/VERSION_PANGOLIN_ALL
-      md5sum: 6c7f688ada8c1e247782758d7fca08e7
+      md5sum: ef9c730f9b45d2fc284575c13fe7c920
     - path: miniwdl_run/call-pangolin4/work/_miniwdl_inputs/0/SRR13687078.ivar.consensus.fasta
       md5sum: d41d8cd98f00b204e9800998ecf8427e
     # sc2 gene coverage
@@ -395,7 +395,7 @@
       md5sum: 60725f5d2663ca8d41955a739a154cc9
     # nextclade
     - path: miniwdl_run/call-nextclade/command
-      md5sum: 2a8dd9cf4d9d3a6b10d034a44c91c4c0
+      md5sum: 25c2a637dcab6e9204192a1fe6386a1b
     - path: miniwdl_run/call-nextclade/inputs.json
       contains: ["dataset_name", "dataset_tag", "genome_fasta"]
     - path: miniwdl_run/call-nextclade/outputs.json
@@ -409,7 +409,7 @@
     - path: miniwdl_run/call-nextclade/work/SRR13687078.ivar.consensus.nextclade.json
     - path: miniwdl_run/call-nextclade/work/SRR13687078.ivar.consensus.nextclade.tsv
     - path: miniwdl_run/call-nextclade/work/NEXTCLADE_VERSION
-      md5sum: d5f4e83525024cba7eb7489cd5209e7d
+      md5sum: 91a455762183b41af0d8de5596e28e7f
     - path: miniwdl_run/call-nextclade/work/nextclade.aligned.fasta
       md5sum: 59893fd3ef32062d50ded18300024734
     - path: miniwdl_run/call-nextclade/work/nextclade.csv

diff --git a/tests/workflows/theiacov/test_wf_theiacov_illumina_se.yml b/tests/workflows/theiacov/test_wf_theiacov_illumina_se.yml
@@ -319,9 +319,9 @@
     - path: miniwdl_run/call-pangolin4/work/PANGOLIN_NOTES
       md5sum: 0b1f8fb5b938fe71631f61234cbf7ab3
     - path: miniwdl_run/call-pangolin4/work/PANGO_ASSIGNMENT_VERSION
-      md5sum: 4f37bb038d973d7e32df5cbbc4067c87
+      md5sum: edfb0c7ebdbf598df19861b650ee904a
     - path: miniwdl_run/call-pangolin4/work/VERSION_PANGOLIN_ALL
-      md5sum: 6c7f688ada8c1e247782758d7fca08e7
+      md5sum: ef9c730f9b45d2fc284575c13fe7c920
     - path: miniwdl_run/call-pangolin4/work/_miniwdl_inputs/0/ERR6319327.ivar.consensus.fasta
       md5sum: d41d8cd98f00b204e9800998ecf8427e
     # sc2 gene coverage
@@ -346,7 +346,7 @@
       md5sum: f7d4aa78c0a111a35f9ed1add781d4dd
     # nextclade
     - path: miniwdl_run/call-nextclade/command
-      md5sum: ad061b04c32b4a3d11ee156684cac457
+      md5sum: ab61d2775776460dc0464393c9e51147
     - path: miniwdl_run/call-nextclade/inputs.json
       contains: ["dataset_name", "dataset_tag", "genome_fasta"]
     - path: miniwdl_run/call-nextclade/outputs.json

diff --git a/tests/workflows/theiacov/test_wf_theiacov_ont.yml b/tests/workflows/theiacov/test_wf_theiacov_ont.yml
@@ -224,9 +224,9 @@
     - path: miniwdl_run/call-pangolin4/work/PANGOLIN_NOTES
       md5sum: 35aa27af5fb90d54561ee9d45a3163d5
     - path: miniwdl_run/call-pangolin4/work/PANGO_ASSIGNMENT_VERSION
-      md5sum: 4f37bb038d973d7e32df5cbbc4067c87
+      md5sum: edfb0c7ebdbf598df19861b650ee904a
     - path: miniwdl_run/call-pangolin4/work/VERSION_PANGOLIN_ALL
-      md5sum: 6c7f688ada8c1e247782758d7fca08e7
+      md5sum: ef9c730f9b45d2fc284575c13fe7c920
     - path: miniwdl_run/call-pangolin4/work/_miniwdl_inputs/0/ont.medaka.consensus.fasta
       md5sum: d41d8cd98f00b204e9800998ecf8427e
     - path: miniwdl_run/call-pangolin4/work/ont.pangolin_report.csv

diff --git a/tests/workflows/theiaprok/test_wf_theiaprok_illumina_pe.yml b/tests/workflows/theiaprok/test_wf_theiaprok_illumina_pe.yml
@@ -221,7 +221,7 @@
     - path: miniwdl_run/call-clean_check_reads/work/_miniwdl_inputs/0/test_1.clean.fastq.gz
     - path: miniwdl_run/call-clean_check_reads/work/_miniwdl_inputs/0/test_2.clean.fastq.gz
     - path: miniwdl_run/call-gambit/command
-      md5sum: bb2310e55a3d567803fab009dd101fe9
+      md5sum: ee3f5f8694857e5de166a905b163e155
     - path: miniwdl_run/call-gambit/inputs.json
       contains: ["assembly", "fasta", "samplename", "test"]
     - path: miniwdl_run/call-gambit/outputs.json
@@ -624,9 +624,9 @@
     - path: miniwdl_run/wdl/tasks/species_typing/task_ts_mlst.wdl
       md5sum: 550791ca5faf11a5f75b6be18739ae01
     - path: miniwdl_run/wdl/tasks/task_versioning.wdl
-      md5sum: 01b9f9c265b15ae776a9fe86efdf59f5
+      md5sum: c178145bad99e3858a1327b3774177f3
     - path: miniwdl_run/wdl/tasks/taxon_id/task_gambit.wdl
-      md5sum: c7f16907ef2bde7ae0e142ae685e19da
+      md5sum: 21bf3c68afcc1c3bc8e70503d77911b7
     - path: miniwdl_run/wdl/tasks/taxon_id/task_kraken2.wdl
       md5sum: f06823a409a087441c8e70e46f7dfc29
     - path: miniwdl_run/wdl/tasks/taxon_id/task_midas.wdl

diff --git a/tests/workflows/theiaprok/test_wf_theiaprok_illumina_se.yml b/tests/workflows/theiaprok/test_wf_theiaprok_illumina_se.yml
@@ -218,7 +218,7 @@
       md5sum: 32c0be4fb7f3030bf9c74c0a836d4f2e
     - path: miniwdl_run/call-clean_check_reads/work/_miniwdl_inputs/0/test_1.clean.fastq.gz
     - path: miniwdl_run/call-gambit/command
-      md5sum: bb2310e55a3d567803fab009dd101fe9
+      md5sum: ee3f5f8694857e5de166a905b163e155
     - path: miniwdl_run/call-gambit/inputs.json
       contains: ["assembly", "fasta", "samplename", "test"]
     - path: miniwdl_run/call-gambit/outputs.json
@@ -592,9 +592,9 @@
     - path: miniwdl_run/wdl/tasks/species_typing/task_ts_mlst.wdl
       md5sum: 550791ca5faf11a5f75b6be18739ae01
     - path: miniwdl_run/wdl/tasks/task_versioning.wdl
-      md5sum: 01b9f9c265b15ae776a9fe86efdf59f5
+      md5sum: c178145bad99e3858a1327b3774177f3
     - path: miniwdl_run/wdl/tasks/taxon_id/task_gambit.wdl
-      md5sum: c7f16907ef2bde7ae0e142ae685e19da
+      md5sum: 21bf3c68afcc1c3bc8e70503d77911b7
     - path: miniwdl_run/wdl/tasks/taxon_id/task_kraken2.wdl
       md5sum: f06823a409a087441c8e70e46f7dfc29
     - path: miniwdl_run/wdl/tasks/taxon_id/task_midas.wdl

diff --git a/workflows/theiacov/wf_theiacov_clearlabs.wdl b/workflows/theiacov/wf_theiacov_clearlabs.wdl
@@ -31,7 +31,7 @@ workflow theiacov_clearlabs {
     File? reference_genome
     # nextclade inputs
     String nextclade_dataset_reference = "MN908947"
-    String nextclade_dataset_tag = "2023-06-16T12:00:00Z"
+    String nextclade_dataset_tag = "2023-08-17T12:00:00Z"
     String? nextclade_dataset_name
     # kraken parameters
     String? target_org

diff --git a/workflows/theiacov/wf_theiacov_fasta.wdl b/workflows/theiacov/wf_theiacov_fasta.wdl
@@ -21,7 +21,7 @@ workflow theiacov_fasta {
     String input_assembly_method
     # nextclade inputs
     String nextclade_dataset_reference = "MN908947"
-    String nextclade_dataset_tag = "2023-06-16T12:00:00Z"
+    String nextclade_dataset_tag = "2023-08-17T12:00:00Z"
     String? nextclade_dataset_name
     # qc check parameters
     File? qc_check_table

diff --git a/workflows/theiacov/wf_theiacov_illumina_pe.wdl b/workflows/theiacov/wf_theiacov_illumina_pe.wdl
@@ -42,9 +42,9 @@ workflow theiacov_illumina_pe {
     Float consensus_min_freq = 0.6 # minimum frequency for a variant to be called as SNP in consensus genome
     Float variant_min_freq = 0.6 # minimum frequency for a variant to be reported in ivar outputs
     # nextclade inputs
-    String nextclade_docker_image = "nextstrain/nextclade:2.13.0"
+    String nextclade_docker_image = "nextstrain/nextclade:2.14.0"
     String nextclade_dataset_reference = "MN908947"
-    String nextclade_dataset_tag = "2023-06-16T12:00:00Z"
+    String nextclade_dataset_tag = "2023-08-17T12:00:00Z"
     String? nextclade_dataset_name
     # nextclade flu inputs
     String nextclade_flu_h1n1_ha_tag = "2023-04-02T12:00:00Z"

diff --git a/workflows/theiacov/wf_theiacov_illumina_se.wdl b/workflows/theiacov/wf_theiacov_illumina_se.wdl
@@ -31,7 +31,7 @@ workflow theiacov_illumina_se {
     Int trim_window_size = 4
     # nextclade inputs
     String nextclade_dataset_reference = "MN908947"
-    String nextclade_dataset_tag = "2023-06-16T12:00:00Z"
+    String nextclade_dataset_tag = "2023-08-17T12:00:00Z"
     String? nextclade_dataset_name
     # reference values
     File? reference_genome

diff --git a/workflows/theiacov/wf_theiacov_ont.wdl b/workflows/theiacov/wf_theiacov_ont.wdl
@@ -31,7 +31,7 @@ workflow theiacov_ont {
     Int min_length = 400
     # nextclade inputs
     String nextclade_dataset_reference = "MN908947"
-    String nextclade_dataset_tag = "2023-06-16T12:00:00Z"
+    String nextclade_dataset_tag = "2023-08-17T12:00:00Z"
     String? nextclade_dataset_name
     # reference values
     File? reference_genome