update submit_genbank

Author: dpark01

pipes/WDL/tasks/tasks_utils.wdl

@@ -1260,3 +1260,29 @@ task filter_sequences_by_length {
         Int  sequences_out     = read_int("OUT_COUNT")
     }
 }
+
+task pair_files_by_basename {
+  input {
+    Array[File] files
+    String      left_ext
+    String      right_ext
+  }
+  command {
+    set -e
+    cp ~{sep=' ' files} .
+  }
+  output {
+    Array[File] left_files  = glob("*.~{left_ext}")
+    Array[File] right_files = glob("*.~{right_ext}")
+    Array[Pair[File,File]] file_pairs = zip(left_files, right_files)
+  }
+  runtime {
+    memory: "1 GB"
+    cpu: 2
+    docker: "ubuntu"
+    disks:  "local-disk 100 HDD"
+    disk: "100 GB" # TES
+    dx_instance_type: "mem1_ssd1_v2_x2"
+    maxRetries: 2
+  }
+}

pipes/WDL/workflows/submit_genbank.wdl

@@ -1,9 +1,9 @@
 version 1.0
 
+import "../tasks/tasks_utils.wdl" as utils
 import "../tasks/tasks_ncbi_tools.wdl" as ncbi_tools
-import "../tasks/tasks_terra.wdl" as terra
 
-workflow submit_sra {
+workflow submit_genbank {
     meta {
         description: "Submit FTP-eligible genomes to NCBI Genbank (currently only flu A/B/C and SARS-CoV-2)"
         author: "Broad Viral Genomics"
@@ -12,30 +12,34 @@ workflow submit_sra {
     }
 
     input {
-        #Array[File]  submission_files
+        Array[File]  submission_files # paired xml and zip files with same basenames
         String       batch_id
 
-        File         genbank_xml
-        File         genbank_zip
-
         File         ncbi_ftp_config_js
         String       prod_test = "Production" # Production or Test
     }
 
     String prefix = "/~{prod_test}/~{batch_id}"
 
-    # TO DO: work in progress, not ready yet
-
-    call ncbi_tools.ncbi_sftp_upload as genbank_upload {
+    call utils.pair_files_by_basename {
         input:
-            config_js        = ncbi_ftp_config_js,
-            submission_xml   = genbank_xml,
-            additional_files = [genbank_zip],
-            target_path      = "~{prefix}/genbank",
-            wait_for         = "1"
+            left_ext = ".xml",
+            right_ext = ".zip",
+            files    = submission_files
+    }
+
+    scatter(file_pair in pair_files_by_basename.file_pairs) {
+        call ncbi_tools.ncbi_sftp_upload as genbank_upload {
+            input:
+                config_js        = ncbi_ftp_config_js,
+                submission_xml   = file_pair.left,
+                additional_files = [file_pair.right],
+                target_path      = "~{prefix}/genbank/~{basename(file_pair.right)}",
+                wait_for         = "1"
+        }
     }
 
     output {
-        Array[File]    genbank_response   = genbank_upload.reports_xmls
+        Array[File]    genbank_response   = flatten(genbank_upload.reports_xmls)
     }
 }