Merge pull request #150 from theiagen/cjk-sra-fetch-defaults

SRA_fetch workflow & `fastq-dl` task improvements
theiagen · Aug 23, 2023 · 4ee9b6d · 4ee9b6d
2 parents 3dfe9ec + 1abb781
commit 4ee9b6d
Show file tree

Hide file tree

Showing 2 changed files with 30 additions and 3 deletions.
diff --git a/tasks/utilities/task_sra_fetch.wdl b/tasks/utilities/task_sra_fetch.wdl
@@ -4,24 +4,46 @@ version 1.0
 task fastq_dl_sra {
   input {
     String sra_accession
-    String docker = "us-docker.pkg.dev/general-theiagen/biocontainers/fastq-dl:2.0.1--pyhdfd78af_0"
+    String docker = "us-docker.pkg.dev/general-theiagen/biocontainers/fastq-dl:2.0.4--pyhdfd78af_0"
     Int disk_size = 100
     Int cpus = 2
     Int memory = 8
-    String? fastq_dl_opts
+    # default set to force the use of SRA instead of ENA due to SRA Lite FASTQ file format issues
+    String fastq_dl_opts = "--provider sra"
+  }
+  meta {
+    # so that call caching is always turned off
+    volatile: true
   }
   command <<<
+    # capture version
     fastq-dl --version | tee VERSION
-    fastq-dl -a ~{sra_accession} ~{fastq_dl_opts}
+
+    # capture date in UTC timezone
+    date -u | tee DATE
+
+    # download fastq files
+    fastq-dl \
+      --verbose \
+      -a ~{sra_accession} \
+      --cpus ~{cpus} \
+      --prefix ~{sra_accession} \
+      ~{fastq_dl_opts}
 
     # tag single-end reads with _1
     if [ -f "~{sra_accession}.fastq.gz" ] && [ ! -f "~{sra_accession}_1.fastq.gz" ]; then
       mv "~{sra_accession}.fastq.gz" "~{sra_accession}_1.fastq.gz"
     fi
+
+
   >>>
   output {
     File read1 = "~{sra_accession}_1.fastq.gz"
     File? read2 = "~{sra_accession}_2.fastq.gz"
+    File fastq_metadata = "~{sra_accession}-run-info.tsv"
+    String fastq_dl_version = read_string("VERSION")
+    String fastq_dl_docker = docker
+    String fastq_dl_date = read_string("DATE")
   }
   runtime {
     docker: docker
@@ -30,5 +52,6 @@ task fastq_dl_sra {
     disks:  "local-disk " + disk_size + " SSD"
     disk: disk_size + " GB" # TES
     preemptible:  1
+    maxRetries: 3
   }
 }
diff --git a/workflows/utilities/data_import/wf_sra_fetch.wdl b/workflows/utilities/data_import/wf_sra_fetch.wdl
@@ -23,5 +23,9 @@ workflow fetch_sra_to_fastq {
   output {
     File read1 = fastq_dl_sra.read1
     File? read2 = fastq_dl_sra.read2
+    File fastq_dl_fastq_metadata = fastq_dl_sra.fastq_metadata
+    String fastq_dl_version = fastq_dl_sra.fastq_dl_version
+    String fastq_dl_docker = fastq_dl_sra.fastq_dl_docker
+    String fastq_dl_date = fastq_dl_sra.fastq_dl_date
   }
 }