update StarAlign.wdl to move chemistry decision logic to the workflow, not the tasks

Author: rsc3

WARP_WDL_Style_Guide.md

@@ -46,6 +46,28 @@
    - Cloud provider support inputs and validation
    - Output block organization
    - Error handling with ErrorWithMessage task
+   - **Workflow-level parameter derivation**
+     - Compute derived values (e.g. tool flags, directory paths, numeric parameters) as WDL expressions in the workflow, not as bash conditionals inside task command blocks
+     - Pass the computed values as explicit task inputs rather than raw enum/mode inputs that the task must interpret
+     - This keeps tasks focused on execution and makes the parameter contract visible in the WDL call site
+     - Example — instead of passing `Int chemistry` and branching in bash:
+       ```wdl
+       # Good: compute in workflow, pass to task
+       Int umi_len = if tenx_chemistry_version == 2 then 10 else 12
+       call MyTask { input: umi_len = umi_len }
+       ```
+       ```wdl
+       # Avoid: passing raw mode flag and branching in bash
+       call MyTask { input: chemistry = tenx_chemistry_version }
+       # ...where the task command block contains if/elif/else to derive umi_len
+       ```
+     - When a derived value requires string parsing that WDL 1.0 cannot express (e.g. parsing a read structure like `"8C18C9M1X"`), use a small dedicated utility task rather than embedding the logic in a large execution task
+   - **Input validation placement**
+     - Validate all user-facing inputs at the workflow level using `ErrorWithMessage` or a dedicated input-checking task (e.g. `checkOptimusInput`), not inside execution tasks
+     - Consolidate validation for a pipeline into a single task or workflow section so there is one place to check for all constraints
+     - Do not duplicate validation in both the workflow and the task — validate once upstream, then trust the inputs downstream
+   - **Derived values with embedded spaces**
+     - When a derived string contains spaces (e.g. `"GeneFull_Ex50pAS Gene"` for STAR's `--soloFeatures`), add a comment at the declaration site noting that the value is intentionally multi-word and must remain unquoted in the command block
 
 ## 6. Task Structure
 
@@ -60,6 +82,11 @@
 - Runtime parameter specification
   - Docker image handling for multi-cloud
   - Memory, CPU, disk sizing
+- **Task purity principle**
+  - Tasks should execute their tool with the parameters they are given, not decide *what* to run based on mode flags
+  - Bash conditionals in command blocks are appropriate for post-execution file handling (moving outputs, renaming files) but not for deriving tool parameters or validating inputs
+  - If a task currently branches on a mode input to select different tool flags, refactor by computing the flag values at the workflow level and passing them as explicit inputs
+  - This makes tasks reusable across workflows without carrying pipeline-specific validation logic
 
 ## 7. Docker Handling
 

pipeline_versions.txt

@@ -11,13 +11,13 @@ ImputationBeagle	 3.0.1	2026-02-23
 JointGenotyping	 1.7.3	2025-08-11 
 MultiSampleSmartSeq2SingleNucleus	 2.2.5	2026-02-24 
 Multiome	 6.1.5	2026-02-24 
-Optimus	 8.0.6	2026-02-24 
+Optimus	 8.0.7	2026-03-31 
 PairedTag	 2.1.11	2026-02-24 
 PeakCalling	 1.0.1	2025-08-11 
 Pipeline Name	Version	Date of Last Commit
 RNAWithUMIsPipeline	 1.0.20	2026-01-21 
 ReblockGVCF	 2.4.4	2026-01-29 
-SlideSeq	 3.6.5	2026-02-24 
+SlideSeq	 3.6.6	2026-03-31 
 SlideTags	 1.0.8	2026-02-24 
 UltimaGenomicsJointGenotyping	 1.2.3	2025-08-11 
 UltimaGenomicsWholeGenomeCramOnly	 1.1.3	2026-01-21 

pipelines/wdl/optimus/Optimus.changelog.md

@@ -1,3 +1,12 @@
+# 8.0.7
+2026-03-31 (Date of Last Commit)
+
+* Moved conditional logic for chemistry, counting mode, strand mode, and solo features out of STARsoloFastq bash command block into workflow-level WDL expressions
+* Added star_strand_mode validation to checkOptimusInput task
+* Replaced chemistry input with explicit umi_len, cb_len, solo_features, and solo_directory task inputs for STARsoloFastq
+* Removed redundant bash input validation from STARsoloFastq task (already validated by checkOptimusInput)
+* No functional changes to pipeline outputs
+
 # 8.0.6
 2026-02-24 (Date of Last Commit)
 

pipelines/wdl/optimus/Optimus.wdl

@@ -78,7 +78,7 @@ workflow Optimus {
   }
 
   # Version of this pipeline
-  String pipeline_version = "8.0.6"
+  String pipeline_version = "8.0.7"
 
   # this is used to scatter matched [r1_fastq, r2_fastq, i1_fastq] arrays
   Array[Int] indices = range(length(r1_fastq))
@@ -148,6 +148,7 @@ workflow Optimus {
       force_no_check = force_no_check,
       counting_mode = counting_mode,
       count_exons = count_exons,
+      star_strand_mode = star_strand_mode,
       gcp_whitelist_v2 = gcp_whitelist_v2,
       gcp_whitelist_v3 = gcp_whitelist_v3,
       azure_whitelist_v2 = azure_whitelist_v2,
@@ -165,14 +166,25 @@ workflow Optimus {
       ubuntu_docker_path = ubuntu_docker_prefix + ubuntu_docker
   }
 
+  # Compute STAR parameters at workflow level instead of bash conditionals in task
+  Int umi_len = if tenx_chemistry_version == 2 then 10 else 12
+  Int cb_len = 16
+  String solo_features = if counting_mode == "sc_rna" then "Gene"
+                         else if count_exons then "GeneFull_Ex50pAS Gene"
+                         else "GeneFull_Ex50pAS"
+  String solo_directory = if counting_mode == "sc_rna" then "Solo.out/Gene" else "Solo.out/GeneFull_Ex50pAS"
+
   call StarAlign.STARsoloFastq as STARsoloFastq {
       input:
         r1_fastq = r1_fastq,
         r2_fastq = r2_fastq,
         star_strand_mode = star_strand_mode,
         white_list = whitelist,
         tar_star_reference = tar_star_reference,
-        chemistry = tenx_chemistry_version,
+        umi_len = umi_len,
+        cb_len = cb_len,
+        solo_features = solo_features,
+        solo_directory = solo_directory,
         counting_mode = counting_mode,
         count_exons = count_exons,
         input_id = input_id,

pipelines/wdl/slideseq/SlideSeq.changelog.md

@@ -1,3 +1,11 @@
+# 3.6.6
+2026-03-31 (Date of Last Commit)
+
+* Moved read structure parsing out of STARsoloFastqSlideSeq bash command block into a new ParseReadStructure task called at the workflow level
+* Moved solo features conditional logic out of STARsoloFastqSlideSeq bash command block into a workflow-level WDL expression
+* Replaced read_structure input with explicit umi_len, cb_len, and solo_features task inputs for STARsoloFastqSlideSeq
+* No functional changes to pipeline outputs
+
 # 3.6.5
 2026-02-24 (Date of Last Commit)
 

pipelines/wdl/slideseq/SlideSeq.wdl

@@ -25,7 +25,7 @@ import "../../../tasks/wdl/Utilities.wdl" as utils
 
 workflow SlideSeq {
 
-    String pipeline_version = "3.6.5"
+    String pipeline_version = "3.6.6"
 
     input {
         Array[File] r1_fastq
@@ -56,6 +56,11 @@ workflow SlideSeq {
     String acr_ubuntu_docker_prefix = "dsppipelinedev.azurecr.io/"
     String ubuntu_docker_prefix = if cloud_provider == "gcp" then gcp_ubuntu_docker_prefix else acr_ubuntu_docker_prefix
 
+    String alpine_docker = "alpine-bash@sha256:965a718a07c700a5204c77e391961edee37477634ce2f9cf652a8e4c2db858ff"
+    String gcp_alpine_docker_prefix = "bashell/"
+    String acr_alpine_docker_prefix = "dsppipelinedev.azurecr.io/"
+    String alpine_docker_prefix = if cloud_provider == "gcp" then gcp_alpine_docker_prefix else acr_alpine_docker_prefix
+
     String gcr_docker_prefix = "us.gcr.io/broad-gotc-prod/"
     String acr_docker_prefix = "dsppipelinedev.azurecr.io/"
 
@@ -100,6 +105,15 @@ workflow SlideSeq {
             sample_id = input_id,
             whitelist = bead_locations
     }
+
+    # Parse read structure into UMI and CB lengths, and compute solo_features at workflow level
+    call StarAlign.ParseReadStructure as ParseReadStructure {
+        input:
+            read_structure = read_structure,
+            alpine_docker_path = alpine_docker_prefix + alpine_docker
+    }
+    String slideseq_solo_features = if count_exons then "Gene GeneFull" else "GeneFull"
+
     scatter(idx in range(length(SplitFastq.fastq_R1_output_array))) {
         call StarAlign.STARsoloFastqSlideSeq as STARsoloFastqSlideSeq {
             input:
@@ -108,7 +122,9 @@ workflow SlideSeq {
                 whitelist = bead_locations,
                 tar_star_reference = tar_star_reference,
                 output_bam_basename = output_bam_basename + "_" + idx,
-                read_structure = read_structure,
+                umi_len = ParseReadStructure.umi_len,
+                cb_len = ParseReadStructure.cb_len,
+                solo_features = slideseq_solo_features,
                 count_exons = count_exons
         }
     }

tasks/wdl/CheckInputs.wdl

@@ -61,6 +61,7 @@ task checkOptimusInput {
     String counting_mode
     Boolean force_no_check
     Boolean count_exons
+    String star_strand_mode
     Int disk = ceil(size(r1_fastq, "GiB")) + 50
     Int machine_mem_mb = 1000
     Int cpu = 1
@@ -99,6 +100,12 @@ task checkOptimusInput {
       echo "ERROR: Invalid value \"${counting_mode}\" for input \"counting_mode\""
     fi
 
+    if [[ ! ("~{star_strand_mode}" == "Forward" || "~{star_strand_mode}" == "Reverse" || "~{star_strand_mode}" == "Unstranded") ]]
+    then
+      pass="false"
+      echo "ERROR: Invalid value \"~{star_strand_mode}\" for input \"star_strand_mode\". Should be Forward, Reverse, or Unstranded."
+    fi
+
     if [[ ~{force_no_check} == "true" ]]
     then
        echo "force_no_check is set: Ignoring input checks"