update pipeline fails, runtime and documentation for kit names

docs/workflows/standalone/dorado_basecalling.md

@@ -27,6 +27,21 @@ Automatic Detection: When set to sup, hac, or fast, Dorado will automatically se
 - `[email protected]`
 - `[email protected]`
 
+??? Info "Supported Kit Names"
+
+    Ensure you use an accepted kit name in the `kit_name` parameter. The following kit names are supported in Dorado:
+
+    ```
+    EXP-NBD103, EXP-NBD104, EXP-NBD114, EXP-NBD114-24, EXP-NBD196, EXP-PBC001, EXP-PBC096, 
+    SQK-16S024, SQK-16S114-24, SQK-LWB001, SQK-MLK111-96-XL, SQK-MLK114-96-XL, SQK-NBD111-24, 
+    SQK-NBD111-96, SQK-NBD114-24, SQK-NBD114-96, SQK-PBK004, SQK-PCB109, SQK-PCB110, SQK-PCB111-24, 
+    SQK-PCB114-24, SQK-RAB201, SQK-RAB204, SQK-RBK001, SQK-RBK004, SQK-RBK110-96, SQK-RBK111-24, 
+    SQK-RBK111-96, SQK-RBK114-24, SQK-RBK114-96, SQK-RLB001, SQK-RPB004, SQK-RPB114-24, 
+    TWIST-16-UDI, TWIST-96A-UDI, VSK-PTC001, VSK-VMK001, VSK-VMK004, VSK-VPS001
+    ```
+
+    Select from these options to avoid input errors.
+
 ### Inputs
 
 | **Terra Task Name** | **Variable** | **Type** | **Description** | **Default Value** | **Terra Status** |

mkdocs.yml

@@ -52,7 +52,7 @@ nav:
                - Zip_Column_Content: workflows/data_export/zip_column_content.md
             - Standalone:
                - Cauris_CladeTyper: workflows/standalone/cauris_cladetyper.md
-               - Dorado_Bsecalling: workflows/standalone/dorado_basecalling.md
+               - Dorado_Basecalling: workflows/standalone/dorado_basecalling.md
                - GAMBIT_Query: workflows/standalone/gambit_query.md
                - Kraken2: workflows/standalone/kraken2.md
                - NCBI-AMRFinderPlus: workflows/standalone/ncbi_amrfinderplus.md

tasks/basecalling/task_dorado_basecall.wdl

@@ -12,7 +12,7 @@ task basecall {
   }
 
   command <<< 
-    set -e
+    set -euo pipefail
 
     # Create output directory for SAM files
     sam_output="output/sam/"
@@ -48,4 +48,6 @@ task basecall {
     disk: disk_size + " GB"
     gpuCount: 1
     gpuType: "nvidia-tesla-t4"  }
+    preemptible: 0
+    maxRetries: 1
 }

tasks/basecalling/task_dorado_demux.wdl

@@ -7,12 +7,12 @@ task dorado_demux {
     String fastq_file_name
     Int cpu = 4 
     Int memory = 16
-    Int disk_size = 50
+    Int disk_size = 100
     String docker = "us-docker.pkg.dev/general-theiagen/staphb/dorado:0.8.0"
   }
 
   command <<< 
-    set -e
+    set -euo pipefail
 
     echo "### Starting Dorado demux ###"
     date
@@ -90,5 +90,7 @@ task dorado_demux {
     cpu: cpu
     memory: "~{memory} GB"
     disks: "local-disk ~{disk_size} SSD"
+    preemptible: 0
+    maxRetries: 1
   }
 }

tasks/basecalling/task_samtools_convert.wdl

@@ -10,7 +10,7 @@ task samtools_convert {
   }
 
   command <<< 
-    set -e
+    set -euo pipefail
 
     bam_output="output/bam/"
     mkdir -p $bam_output
@@ -20,7 +20,7 @@ task samtools_convert {
       bam_file="${bam_output}/${base_name}.bam"
 
       echo "Converting SAM to BAM: $sam_file -> $bam_file"
-      samtools view -bS "$sam_file" > "$bam_file" || { echo "ERROR: samtools failed to convert $sam_file to BAM" >&2; exit 1; }
+      samtools view -@ ~{cpu} -bS "$sam_file" > "$bam_file" || { echo "ERROR: samtools failed to convert $sam_file to BAM" >&2; exit 1; }
 
       echo "Conversion successful: $bam_file"
     done
@@ -36,6 +36,6 @@ task samtools_convert {
     docker: docker
     cpu: cpu
     memory: "~{memory} GB"
-    disks: "local-disk ~{disk_size} HDD"
+    disks: "local-disk ~{disk_size} SSD"
   }
 }

workflows/utilities/data_import/wf_update_srr_metadata.wdl

@@ -0,0 +1,22 @@
+version 1.0
+
+import "../../../tasks/utilities/data_handling/task_fetch_srr_metadata.wdl" as srr_task
+
+workflow wf_retrieve_srr {
+  meta {
+    description: "This workflow retrieves the Sequence Read Archive (SRA) accession (SRR) associated with a given sample accession. It uses the fastq-dl tool to fetch metadata from SRA and outputs the SRR accession that can be used for downstream analysis."
+  }
+  input {
+    String sample_accession  
+    String retrieve_srr_docker = "us-docker.pkg.dev/general-theiagen/biocontainers/fastq-dl:2.0.4--pyhdfd78af_0"
+  }
+
+  call srr_task.fetch_srr_metadata {
+    input:
+      sample_accession = sample_accession,  
+      docker = retrieve_srr_docker
+  }
+  output {
+    String srr_accession = fetch_srr_metadata.srr_accession 
+  }
+}