From 844a5142c6d4b2bd6e4df252efb2fb49d41427cf Mon Sep 17 00:00:00 2001
From: fraser-combe <frascom@hotmail.com>
Date: Wed, 18 Dec 2024 08:12:22 -0600
Subject: [PATCH] pass file paths to basecall

---
 tasks/basecalling/task_dorado_basecall.wdl      |  3 ---
 .../file_handling/task_transfer_pod5_files.wdl  | 17 +++--------------
 workflows/utilities/wf_dorado_basecalling.wdl   |  1 +
 3 files changed, 4 insertions(+), 17 deletions(-)

diff --git a/tasks/basecalling/task_dorado_basecall.wdl b/tasks/basecalling/task_dorado_basecall.wdl
index 804399998..1d79ce274 100644
--- a/tasks/basecalling/task_dorado_basecall.wdl
+++ b/tasks/basecalling/task_dorado_basecall.wdl
@@ -10,7 +10,6 @@ task basecall {
     Int cpu = 8
     String docker = "us-docker.pkg.dev/general-theiagen/staphb/dorado:0.8.3"
   }
-
   command <<< 
     set -euo pipefail
 
@@ -69,7 +68,6 @@ task basecall {
 
     echo "Basecalling completed for ~{input_file}. SAM file renamed to: $sam_file" | tee -a "dorado_basecall.log"
   >>>
-  
   output {
     Array[File] sam_files = glob("output/sam_*/*.sam")
     String dorado_docker = docker
@@ -78,7 +76,6 @@ task basecall {
     # keeping this dorado_log just for debugging purposes, not a wf output
     File dorado_log = "dorado_basecall.log"
   }
-  
   runtime {
     docker: docker
     cpu: cpu
diff --git a/tasks/utilities/file_handling/task_transfer_pod5_files.wdl b/tasks/utilities/file_handling/task_transfer_pod5_files.wdl
index c0fd8c8dd..07326d5e7 100644
--- a/tasks/utilities/file_handling/task_transfer_pod5_files.wdl
+++ b/tasks/utilities/file_handling/task_transfer_pod5_files.wdl
@@ -2,7 +2,7 @@ version 1.0
 
 task transfer_pod5_files {
   input {
-    String pod5_bucket_path             # Terra bucket path (e.g., "gs://your-terra-bucket/pod5_uploads/")
+    String pod5_bucket_path             # GCS bucket path containing `.pod5` files (e.g., "gs://your-terra-bucket/pod5_uploads/")
     Int disk_size = 100
     Int memory = 32
     Int cpu = 8
@@ -11,10 +11,7 @@ task transfer_pod5_files {
   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}"
+     echo "Listing .pod5 files in ~{pod5_bucket_path}"
     gcloud storage ls -r "~{pod5_bucket_path}" | grep "\.pod5$" > pod5_files_list.txt
 
     # Check if any files are found
@@ -22,17 +19,9 @@ task transfer_pod5_files {
       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
+    Array[File] pod5_file_paths = read_lines("pod5_files_list.txt")
   }
    runtime {
     docker: docker
diff --git a/workflows/utilities/wf_dorado_basecalling.wdl b/workflows/utilities/wf_dorado_basecalling.wdl
index 1c2fe3487..33da91b1b 100644
--- a/workflows/utilities/wf_dorado_basecalling.wdl
+++ b/workflows/utilities/wf_dorado_basecalling.wdl
@@ -28,6 +28,7 @@ workflow dorado_basecalling_workflow {
   call versioning_task.version_capture {
     input:
   }
+  # List `.pod5` files in GCS bucket
   call transfer_pod5_files_task.transfer_pod5_files as transfer_pod5 {
     input:
       pod5_bucket_path = pod5_bucket_path