handle multiple SRR accessionss as string version outputs

docs/workflows/public_data_sharing/fetch_srr_accession.md

@@ -8,7 +8,7 @@
 
 ## Fetch SRR Accession
 
-This workflow is designed to retrieve the Sequence Read Archive (SRA) accession (SRR) associated with a given sample accession. The primary inputs are BioSample IDs (e.g., SAMN00000000) or SRA Experiment IDs (e.g., SRX000000), which link to sequencing data in the SRA repository.
+This workflow retrieves the Sequence Read Archive (SRA) accession (SRR) associated with a given sample accession. The primary inputs are BioSample IDs (e.g., SAMN00000000) or SRA Experiment IDs (e.g., SRX000000), which link to sequencing data in the SRA repository.
 
 The workflow uses the fastq-dl tool to fetch metadata from SRA and specifically parses this metadata to extract the associated SRR accession and outputs the SRR accession.
 

mkdocs.yml

@@ -43,8 +43,8 @@ nav:
                - Samples_to_Ref_Tree: workflows/phylogenetic_placement/samples_to_ref_tree.md
                - Usher_PHB: workflows/phylogenetic_placement/usher.md
             - Public Data Sharing:
+               - Fetch_SRR_Accession: workflows/public_data_sharing/fetch_srr_accessions.md
                - Mercury_Prep_N_Batch: workflows/public_data_sharing/mercury_prep_n_batch.md
-               - Retrieve_SRR_Metadata: workflows/public_data_sharing/retrieve_srr_metadata.md
                - Terra_2_GISAID: workflows/public_data_sharing/terra_2_gisaid.md
                - Terra_2_NCBI: workflows/public_data_sharing/terra_2_ncbi.md
             - Exporting Data from Terra:

tasks/utilities/data_handling/task_fetch_srr_accession.wdl

@@ -10,39 +10,43 @@ task fetch_srr_accession {
   }
   meta {
     volatile: true
+    version: "1.0"
   }
 
   command <<< 
-    mkdir -p metadata_output
     date -u | tee DATE
 
+    fastq-dl --version | tee VERSION
+
     # Debug output to show the sample being processed
     echo "Fetching metadata for sample accession: ${sample_accession}"
 
-    # Use fastq-dl to fetch metadata only
-    fastq-dl --accession ~{sample_accession} --outdir metadata_output --only-download-metadata --verbose
-
+    # Use fastq-dl to fetch metadata only, outputting to the current directory
+    fastq-dl --accession ~{sample_accession} --only-download-metadata --verbose
 
-    if [[ -f metadata_output/fastq-run-info.tsv ]]; then
+    if [[ -f fastq-run-info.tsv ]]; then
       echo "Metadata written for ${sample_accession}:"
       echo "TSV content:"
-      cat metadata_output/fastq-run-info.tsv
+      cat fastq-run-info.tsv
+
+      # Extract the SRR accessions and write them directly to srr_accession.txt
+      awk -F'\t' 'NR>1 {print $1}' fastq-run-info.tsv > srr_accession.txt
 
-      # Extract the SRR accession (It is typically in the first column)
-      SRR_accessions=$(awk -F'\t' 'NR>1 {print $1}' metadata_output/fastq-run-info.tsv)
-      if [[ -z "${SRR_accessions}" ]]; then
-        echo "No SRR accession found for ${sample_accession}" > metadata_output/srr_accession.txt
+      # Check if srr_accession.txt is empty (no SRR accessions found)
+      if [[ ! -s srr_accession.txt ]]; then
+        echo "No SRR accession found for ${sample_accession}" > srr_accession.txt
       else
-        echo "Extracted SRR accessions: ${SRR_accessions}"
-        echo "${SRR_accessions}" > metadata_output/srr_accession.txt
+        echo "Extracted SRR accessions:"
+        cat srr_accession.txt
       fi
     else
       echo "No metadata found for ${sample_accession}"
-      echo "No SRR accession found" > metadata_output/srr_accession.txt
+      echo "No SRR accession found" > srr_accession.txt
     fi
   >>>
   output {
-    String srr_accession = read_string("metadata_output/srr_accession.txt")
+    Array[String] srr_accession = read_lines("srr_accession.txt")
+    String fastq_dl_version = read_string("VERSION")
   }
   runtime {
     docker: docker

workflows/utilities/data_import/wf_fetch_srr_accession.wdl

@@ -4,16 +4,23 @@ import "../../../tasks/utilities/data_handling/fetch_srr_accession.wdl" as srr_t
 
 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."
+    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."
   }
   input {
     String sample_accession  
   }
-  call srr_task.fetch_srr_metadata {
+  call versioning_task.version_capture {
+    input:
+  }
+  call srr_task.fetch_srr_accession {
     input:
       sample_accession = sample_accession
   }
   output {
-    String srr_accession = fetch_srr_metadata.srr_accession 
+    Array[String] srr_accession = fetch_srr_accession.srr_accession
+
+    # Version Captures
+    String phb_version = version_capture.phb_version
+    String fetch_srr_date = version_capture.date
   }
 }