fix file extension awareness cg_pipeline (#189)

* fix file extension awareness cg_pipeline

* remove legacy task files

* add back taxonID and versioning

* Update runtime for Azure compatibility

Co-authored-by: kevinlibuit <kevin@libuit.com>

Author: michellescribner

tasks/assembly/task_mycosnp_consensus_assembly.wdl

@@ -10,6 +10,7 @@ task mycosnp {
     String accession = "GCA_016772135"
     Int memory = 16
     Int cpu = 4
+    Int disk_size = 100
   }
   command <<<
     date | tee DATE
@@ -45,7 +46,8 @@ task mycosnp {
     docker: "~{docker}"
     memory: "~{memory} GB"
     cpu: cpu
-    disks:  "local-disk 50 SSD"
+    disks:  "local-disk " + disk_size + " SSD"
+    disk: disk_size + " GB"
     maxRetries: 3
     preemptible: 0
   }

tasks/assembly/task_shovill.wdl

@@ -6,6 +6,7 @@ task shovill_pe {
     File read2_cleaned
     String samplename
     String docker = "quay.io/staphb/shovill:1.1.0"
+    Int disk_size = 100
 
     ## SHOVILL optional parameters
     ##  --depth [INT]           Sub-sample --R1/--R2 to this depth. Disable with --depth 0 (default: 150)
@@ -70,11 +71,13 @@ task shovill_pe {
     String shovill_version = read_string("VERSION")
   }
   runtime {
-      docker: "~{docker}"
-      memory: "16 GB"
-      cpu: 4
-      disks: "local-disk 100 SSD"
-      preemptible: 0
+    docker: "~{docker}"
+    memory: "16 GB"
+    cpu: 4
+    disks:  "local-disk " + disk_size + " SSD"
+    disk: disk_size + " GB"
+    maxRetries: 3
+    preemptible: 0
   }
 }
 
@@ -83,6 +86,7 @@ task shovill_se {
     File read1_cleaned
     String samplename
     String docker = "quay.io/staphb/shovill-se:1.1.0"
+    Int disk_size = 100
 
     ## SHOVILL optional parameters
     ##  --depth [INT]           Sub-sample --R1/--R2 to this depth. Disable with --depth 0 (default: 150)
@@ -141,10 +145,12 @@ task shovill_se {
     String shovill_version = read_string("VERSION")
   }
   runtime {
-      docker: "~{docker}"
-      memory: "16 GB"
-      cpu: 4
-      disks: "local-disk 100 SSD"
-      preemptible: 0
+    docker: "~{docker}"
+    memory: "16 GB"
+    cpu: 4
+    disks:  "local-disk " + disk_size + " SSD"
+    disk: disk_size + " GB"
+    maxRetries: 3
+    preemptible: 0
   }
 }

tasks/gene_typing/task_abricate.wdl

@@ -12,6 +12,7 @@ task abricate {
     Int? mincov
     Int cpu = 2
     String docker = "staphb/abricate:1.0.1-abaum-plasmid"
+    Int disk_size = 100
   }
   command <<<
     date | tee DATE
@@ -49,6 +50,8 @@ task abricate {
     memory: "8 GB"
     cpu: cpu
     docker: docker
-    disks: "local-disk 100 HDD"
+    disks:  "local-disk " + disk_size + " SSD"
+    disk: disk_size + " GB"
+    maxRetries: 3
   }
 }

tasks/gene_typing/task_amrfinderplus.wdl

@@ -13,6 +13,7 @@ task amrfinderplus_nuc {
     Boolean detailed_drug_class = false
     Int cpu = 4
     String docker = "staphb/ncbi-amrfinderplus:3.10.42"
+    Int disk_size = 100
     Boolean hide_point_mutations = false
   }
   command <<<
@@ -163,7 +164,8 @@ task amrfinderplus_nuc {
     memory: "8 GB"
     cpu: cpu
     docker: docker
-    disks: "local-disk 100 SSD"
+    disks:  "local-disk " + disk_size + " SSD"
+    disk: disk_size + " GB"
     preemptible: 0
     maxRetries: 3
   }

tasks/gene_typing/task_bakta.wdl

@@ -8,6 +8,7 @@ task bakta {
     Int cpu = 8
     Int memory = 16
     String docker = "quay.io/biocontainers/bakta:1.5.1--pyhdfd78af_0"
+    Int disk_size = 100
     # Parameters 
     #  proteins: Fasta file of trusted protein sequences for CDS annotation
     #  prodigal_tf: Prodigal training file to use for CDS prediction
@@ -61,6 +62,8 @@ task bakta {
     memory: "~{memory} GB"
     cpu: cpu
     docker: docker
-    disks: "local-disk 100 SSD"
+    disks:  "local-disk " + disk_size + " SSD"
+    disk: disk_size + " GB"
+    maxRetries: 3
   }
 }

tasks/gene_typing/task_gamma.wdl

@@ -6,6 +6,7 @@ task gamma_one_sample {
     String samplename
     String docker = "quay.io/biocontainers/gamma:1.4--hdfd78af_0"
     File gamma_database
+    Int disk_size = 100
   }
   String database_name = basename(gamma_database)
   command <<<
@@ -23,6 +24,8 @@ task gamma_one_sample {
     memory: "8 GB"
     cpu: 4
     docker: "~{docker}"
-    disks: "local-disk 100 HDD"
+    disks: "local-disk " + disk_size + " SSD"
+    disk: disk_size + " GB"
+    maxRetries: 3
   }
 }

tasks/gene_typing/task_plasmidfinder.wdl

@@ -7,6 +7,7 @@ task plasmidfinder {
     Int cpu = 8
     Int memory = 16
     String docker = "staphb/plasmidfinder:2.1.6"
+    Int disk_size = 100
     String? database
     String? database_path
     String? method_path
@@ -63,6 +64,8 @@ task plasmidfinder {
     memory: "~{memory} GB"
     cpu: cpu
     docker: "~{docker}"
-    disks: "local-disk 100 HDD"
+    disks: "local-disk " + disk_size + " SSD"
+    disk: disk_size + " GB"
+    maxRetries: 3
   }
 }

tasks/gene_typing/task_prokka.wdl

@@ -7,6 +7,7 @@ task prokka {
     Int cpu = 8
     Int memory = 16
     String docker = "staphb/prokka:1.14.5"
+    Int disk_size = 100
     # Parameters 
     #  proteins recommended: when you have good quality reference genomes and want to ensure gene naming is consistent [false]
     #  prodigal_tf: prodigal training file
@@ -42,6 +43,8 @@ task prokka {
     memory: "~{memory} GB"
     cpu: cpu
     docker: docker
-    disks: "local-disk 100 HDD"
+    disks: "local-disk " + disk_size + " SSD"
+    disk: disk_size + " GB"
+    maxRetries: 3
   }
 }

tasks/gene_typing/task_resfinder.wdl

@@ -10,7 +10,7 @@ task resfinder {
     Float? threshold = 0.9 # Threshold for identity of ResFinder
     Boolean point = false # Run pointfinder for chromosomal mutations
     String docker = "staphb/resfinder:4.1.11"
-
+    Int disk_size = 100
   }
   command <<<
     date | tee DATE
@@ -100,6 +100,8 @@ task resfinder {
     memory: "8 GB"
     cpu: 4
     docker: docker
-    disks: "local-disk 100 HDD"
+    disks: "local-disk " + disk_size + " SSD"
+    disk: disk_size + " GB"
+    maxRetries: 3
   }
 }

tasks/phylogenetic_inference/task_iqtree.wdl

@@ -9,6 +9,7 @@ task iqtree {
     String alrt = 1000 # SH-like approximate likelihood ratio test (SH-aLRT) replicates
     String? iqtree_opts = ""
     String docker = "staphb/iqtree:1.6.7"
+    Int disk_size = 100
   }
   command <<<
     # date and version control
@@ -39,7 +40,8 @@ task iqtree {
     docker: "~{docker}"
     memory: "32 GB"
     cpu: 4
-    disks: "local-disk 100 SSD"
+    disks: "local-disk " + disk_size + " SSD"
+    disk: disk_size + " GB"
     preemptible: 0
     maxRetries: 3
   }