PD-2759 Add positioning task to Slide-tags WDL (#1402)

broadinstitute · Nov 5, 2024 · f81cee2 · f81cee2
1 parent 70d3e1d
commit f81cee2
Show file tree

Hide file tree

Showing 4 changed files with 109 additions and 12 deletions.
diff --git a/beta-pipelines/skylab/slidetags/SlideTags.wdl b/beta-pipelines/skylab/slidetags/SlideTags.wdl
@@ -1,6 +1,7 @@
 version 1.0
 
 import "scripts/spatial-count.wdl" as SpatialCount
+import "scripts/positioning.wdl" as Positioning
 
 workflow SlideTags {
 
@@ -10,25 +11,28 @@ workflow SlideTags {
         String id
         Array[String] fastq_paths
         Array[String] pucks
-        Int mem_GiB = 64
-        Int disk_GiB = 128
-        String docker = "us.gcr.io/broad-gotc-prod/slide-tags:1.0.0"
+        Array[String] rna_paths
+        String sb_path
+        String docker = "us.gcr.io/broad-gotc-prod/slide-tags:1.1.0"
      }
 
     parameter_meta {
         fastq_paths: "Array of paths to spatial fastq files"
         pucks: "Array of paths to puck files"
-        mem_GiB: "Memory in GiB to allocate to the task"
-        disk_GiB: "Disk in GiB to allocate to the task"
         docker: "Docker image to use"
     }
 
     call SpatialCount.count as spatial_count {
         input:
             fastq_paths = fastq_paths,
             pucks = pucks,
-            mem_GiB = mem_GiB,
-            disk_GiB = disk_GiB,
+            docker = docker
+     }
+
+    call Positioning.generate_positioning as positioning {
+        input:
+            rna_paths = rna_paths,
+            sb_path = spatial_count.sb_counts,
             docker = docker
      }
 

diff --git a/beta-pipelines/skylab/slidetags/scripts/CREDITS.md b/beta-pipelines/skylab/slidetags/scripts/CREDITS.md
@@ -6,6 +6,10 @@ This project uses code from the following sources:
   URL: [https://github.com/MacoskoLab/Macosko-Pipelines/blob/main/spatial-count/spatial-count.wdl](https://github.com/MacoskoLab/Macosko-Pipelines/blob/main/spatial-count/spatial-count.wdl)  
   This code was adapted and modified from the Macosko Lab's pipeline repository.
 
+- **Positioning Workflow**
+  URL: [https://github.com/MacoskoLab/Macosko-Pipelines/blob/main/positioning/positioning.wdl](https://github.com/MacoskoLab/Macosko-Pipelines/blob/main/positioning/positioning.wdl)  
+  This code was adapted and modified from the Macosko Lab's pipeline repository.
+
 Additional modifications include output handling and script download changes.
 
 Please refer to the original source for the full context of the workflow.
diff --git a/beta-pipelines/skylab/slidetags/scripts/positioning.wdl b/beta-pipelines/skylab/slidetags/scripts/positioning.wdl
@@ -0,0 +1,90 @@
+version 1.0
+
+task generate_positioning {
+  input {
+    Array[String] rna_paths
+    String sb_path
+    Int mem_GiB  = 128
+    Int disk_GiB = 128
+    Int nthreads = 16
+    String docker
+  }
+  command <<<
+    set -euo pipefail
+    set -x
+    echo "<< starting spatial-count >>"
+
+    gcloud config set storage/process_count 16 # is this set by user?
+    gcloud config set storage/thread_count  2 # is this set by user?
+
+    # Download the scripts -- these need to be changed -- also need to add to docker   
+    wget https://raw.githubusercontent.com/MacoskoLab/Macosko-Pipelines/ee005109446f58764509ee47ff51c212ce8dabe3/positioning/positioning.R
+    wget https://raw.githubusercontent.com/MacoskoLab/Macosko-Pipelines/6a78716aa08a9f2506c06844f7e3fd491b03aa8b/positioning/load_matrix.R
+    wget https://raw.githubusercontent.com/MacoskoLab/Macosko-Pipelines/a7fc86abbdd3d46461c500e7d024315d88a97e9a/positioning/run-positioning.R
+
+    echo "RNA: ~{sep=' ' rna_paths}"
+    echo "SB: ~{sb_path}"
+
+    # Assert that the RNA files exist
+    rnas=(~{sep=' ' rna_paths})
+    for rna in "${rnas[@]}" ; do
+        if ! gsutil stat "$rna" &> /dev/null ; then
+            echo "ERROR: gsutil stat command failed on file $rna"
+            exit 1
+        fi
+    done
+
+    # Download the RNA
+    echo "Downloading RNA:"
+    mkdir RNA
+    gcloud storage cp ~{sep=' ' rna_paths} RNA
+
+    # Assert that the SB file exists
+    if ! gsutil stat "~{sb_path}" &> /dev/null ; then
+        echo "ERROR: gsutil stat command failed on file ~{sb_path}"
+        exit 1
+    fi
+
+    # Download the SB
+    echo "Downloading SB:"
+    mkdir SB
+    gcloud storage cp ~{sb_path} SB
+
+    # Run the script
+    echo ; echo "Running run-positioning.R"
+    Rscript run-positioning.R RNA SB output
+
+    # Upload the results
+    ls output/* 
+
+    if [[ -f output/seurat.qs ]] ; then
+        echo "true" > DONE
+    else
+        echo ; echo "ERROR: CANNOT FIND: seurat.qs"
+    fi
+
+    echo; echo "Writing logs:"
+    echo; echo "RNA size:"; du -sh RNA
+    echo; echo "SB size:"; du -sh SB
+    echo; echo "output size:"; du -sh output
+    echo; echo "FREE SPACE:"; df -h
+
+    echo "tar files/logs"
+    cat stdout stderr > positioning.log
+    tar -zcvf output.tar.gz output
+    echo "<< completed positioning >>"
+  >>>
+
+  output {
+    File output_file = "output.tar.gz"
+    File positioning_log = "positioning.log"
+  }
+
+  runtime {
+    docker: docker
+    memory: "~{mem_GiB} GB"
+    disks: "local-disk ~{disk_GiB} SSD"
+    cpu: nthreads
+  }
+
+}
diff --git a/beta-pipelines/skylab/slidetags/scripts/spatial-count.wdl b/beta-pipelines/skylab/slidetags/scripts/spatial-count.wdl
@@ -4,8 +4,9 @@ task count {
   input {
     Array[String] fastq_paths
     Array[String] pucks
-    Int mem_GiB
-    Int disk_GiB
+    Int mem_GiB = 64
+    Int disk_GiB = 128
+    Int nthreads = 1
     String docker
   }
   command <<<
@@ -74,7 +75,6 @@ task count {
   >>>
 
   output {
-    Boolean DONE = read_boolean("DONE")
     File sb_counts = "SBcounts.h5"
     File spatial_log = "spatial-count.log"
 
@@ -83,8 +83,7 @@ task count {
     docker: docker
     memory: "~{mem_GiB} GB"
     disks: "local-disk ~{disk_GiB} SSD"
-    cpu: 1
-    preemptible: 0
+    cpu: nthreads
   }
 }