Add and use disk_space param for stage_files to avoid running out of space (366) (#89)

Author: David Shiga

adapter_pipelines/cellranger/adapter.wdl

@@ -256,6 +256,9 @@ workflow Adapter10xCount {
       record_http = record_http,
       pipeline_tools_version = pipeline_tools_version,
       add_md5s = add_md5s,
-      pipeline_version = analysis.pipeline_version
+      pipeline_version = analysis.pipeline_version,
+      # The sorted bam is the largest output. Other outputs will increase space by ~50%.
+      # Factor of 2 and addition of 50 GB gives some buffer.
+      disk_space = ceil(size(analysis.sorted_bam, "GB") * 2 + 50)
   }
 }

adapter_pipelines/ss2_single_sample/adapter.wdl

@@ -210,6 +210,8 @@ workflow AdapterSmartSeq2SingleCell{
       pipeline_tools_version = pipeline_tools_version,
       add_md5s = add_md5s,
       max_retries = max_cromwell_retries,
-      pipeline_version = analysis.pipeline_version
+      pipeline_version = analysis.pipeline_version,
+      # The bam files are by far the largest outputs. The extra 5 GB should easily cover everything else.
+      disk_space = ceil(size(analysis.aligned_bam, "GB") + size(analysis.aligned_transcriptome_bam, "GB") + 5)
   }
 }

adapter_pipelines/submit.wdl

@@ -133,6 +133,7 @@ task stage_files {
   Boolean record_http
   String pipeline_tools_version
   Int max_retries = 0
+  Int disk_space
 
   command <<<
     set -e
@@ -177,6 +178,7 @@ task stage_files {
 
   runtime {
     docker: "quay.io/humancellatlas/secondary-analysis-pipeline-tools:" + pipeline_tools_version
+    disks: "local-disk ${disk_space} HDD"
     maxRetries: max_retries
   }
   output {
@@ -253,6 +255,8 @@ workflow submit {
   Int max_retries = 0
   # Version of the pipeline, should be included in the pipeline file
   String pipeline_version
+  # Disk space to allocate for stage_files task
+  Int disk_space
 
   call get_metadata {
     input:
@@ -304,7 +308,8 @@ workflow submit {
       retry_max_interval = retry_max_interval,
       record_http = record_http,
       pipeline_tools_version = pipeline_tools_version,
-      max_retries = max_retries
+      max_retries = max_retries,
+      disk_space = disk_space
   }
 
   call confirm_submission {