new tranfer from bucket task

tasks/basecalling/task_dorado_basecall.wdl

@@ -8,7 +8,7 @@ task basecall {
     Int disk_size = 100
     Int memory = 32
     Int cpu = 8
-    String docker = "us-docker.pkg.dev/general-theiagen/staphb/dorado:0.8.0"
+    String docker = "us-docker.pkg.dev/general-theiagen/staphb/dorado:0.8.3"
   }
 
   command <<< 

tasks/utilities/file_handling/task_transfer_pod5_files.wdl

@@ -0,0 +1,45 @@
+version 1.0
+
+task transfer_pod5_files {
+  input {
+    String pod5_bucket_path             # Terra bucket path (e.g., "gs://your-terra-bucket/pod5_uploads/")
+    Int disk_size = 100
+    Int memory = 32
+    Int cpu = 8
+    String docker =  "us-docker.pkg.dev/general-theiagen/cloudsdktool/google-cloud-cli:427.0.0-alpine"
+  }
+  command <<<
+     set -euo pipefail
+
+     # Create a directory for downloaded `.pod5` files
+    mkdir -p pod5_downloads
+
+    echo "Listing and downloading .pod5 files from ~{pod5_bucket_path}"
+    gcloud storage ls -r "~{pod5_bucket_path}" | grep "\.pod5$" > pod5_files_list.txt
+
+    # Check if any files are found
+    if [ ! -s pod5_files_list.txt ]; then
+      echo "ERROR: No POD5 files found in ~{pod5_bucket_path}" >&2
+      exit 1
+    fi
+
+    # Download all `.pod5` files locally
+    while read -r file_path; do
+      local_path="pod5_downloads/$(basename "$file_path")"
+      gcloud storage cp "$file_path" "$local_path" || { echo "ERROR: Failed to download $file_path"; exit 1; }
+      echo "$local_path" >> downloaded_pod5_files.txt
+    done < pod5_files_list.txt
+  >>>
+
+  output {
+    Array[File] pod5_file_paths = read_lines("downloaded_pod5_files.txt")  # Local paths of downloaded `.pod5` files
+  }
+   runtime {
+    docker: docker
+    cpu: cpu
+    memory: "~{memory} GB"
+    disks: "local-disk ~{disk_size} SSD"
+    preemptible: 0
+    maxRetries: 1
+  }
+}

workflows/utilities/wf_dorado_basecalling.wdl

@@ -6,13 +6,14 @@ import "../../tasks/basecalling/task_dorado_demux.wdl" as dorado_demux_task
 import "../../tasks/utilities/file_handling/task_transfer_files.wdl" as transfer_fastq_files
 import "../../tasks/utilities/data_import/task_create_terra_table.wdl" as terra_fastq_table
 import "../../tasks/task_versioning.wdl" as versioning_task
+import "../../tasks/utilities/file_handling/task_transfer_pod5_files.wdl" as transfer_pod5_files_task         
 
 workflow dorado_basecalling_workflow {
   meta {
     description: "GPU-accelerated workflow for basecalling Oxford Nanopore POD5 files, generating SAM outputs and supporting downstream demultiplexing and FASTQ output."
   }
   input {
-    Array[File] input_files
+    String pod5_bucket_path  # GCS bucket path containing POD5 files
     String dorado_model = "sup" 
     String kit_name
     String new_table_name
@@ -27,12 +28,16 @@ workflow dorado_basecalling_workflow {
   call versioning_task.version_capture {
     input:
   }
-  scatter (file in input_files) {
+  call transfer_pod5_files_task.transfer_pod5_files as transfer_pod5 {
+    input:
+      pod5_bucket_path = pod5_bucket_path
+  }
+  scatter (pod5_path in transfer_pod5.pod5_file_paths) {
     call basecall_task.basecall as dorado_basecall {
       input:
-        input_file = file,
+        input_file = pod5_path,
         dorado_model = dorado_model,
-        kit_name = kit_name,
+        kit_name = kit_name
     }
   }
   call samtools_convert_task.samtools_convert {