Feature/full run

.github/pull_request_template.md

@@ -0,0 +1,28 @@
+## Description
+
+
+## Dependencies (Issues/PRs)
+
+
+## Type of Change
+- [ ] Bug fix
+- [ ] New feature
+- [ ] Documentation
+
+## Task Checklist
+- [ ] documentation updated
+- [ ] `miniwdl check` passed
+- [ ] `womtool validate` passed
+
+
+## Testing
+### Workflow Engine Tested On
+- [ ] HealthOmics, Amazon Web Services
+- [ ] Google Cloud Platform, Cromwell
+- [ ] Google Cloud Platform
+- [ ] Microsoft Azure, Cromwell
+- [ ] GA4GH Workflow Execution Service, On Premises and Multi Cloud
+- [ ] Other
+
+### Successful Workflow IDs
+*

.github/workflows/lint-test-workflows.yml

@@ -21,5 +21,5 @@ jobs:
           workbench-workflow-service-url:  ${{ secrets.WORKBENCH_WORKFLOW_SERVICE_URL }}
           workbench-ewes-refresh-token: ${{ secrets.WORKBENCH_EWES_REFRESH_TOKEN }}
           workbench-workflow-service-refresh-token: ${{ secrets.WORKBENCH_WORKFLOW_SERVICE_REFRESH_TOKEN }}
-          wdl-ci-custom-test-wdl-dir: pmdbs-spatial-transcriptomics-wdl-ci-custom-test-dir
+          wdl-ci-custom-test-wdl-dir: spatial-transcriptomics-wdl-ci-custom-test-dir
 

docker/spatial_py/scripts/geomx_merge_and_prep

@@ -48,7 +48,7 @@ def main(args):
     hvg_features = hvg_adata.var.copy()
 
     sc.pl.highly_variable_genes(
-        hvg_adata,
+        merged_adata,
     )
     fig = plt.gcf()
     fig.suptitle(f"Highly variable genes dispersion plot - {args.output_prefix}", va="center", ha="center", fontsize=16)

docker/spatial_py/scripts/integrate_harmony

@@ -15,10 +15,21 @@ def main(args):
     ##################################
     adata = sc.read_h5ad(args.adata_input)
 
-    sc.external.pp.harmony_integrate(
-        adata,
-        key=args.batch_key,
-    )
+    n_cells = adata.n_obs
+    if n_cells < 100:
+        nclust = min(10, max(3, n_cells // 5))
+        print(f"[INFO] Detected small dataset (n={n_cells}). Setting nclust={nclust}.")
+        sc.external.pp.harmony_integrate(
+            adata,
+            key=args.batch_key,
+            nclust=nclust,
+        )
+    else:
+        print(f"[INFO] Running Harmony with default settings (n={n_cells}).")
+        sc.external.pp.harmony_integrate(
+            adata,
+            key=args.batch_key,
+        )
 
     # Save outputs
     adata.write_h5ad(filename=args.adata_output, compression="gzip")

docker/spatial_py/scripts/visium_process

@@ -62,7 +62,7 @@ def main(args):
     hvg_features = hvg_adata.var.copy()
 
     sc.pl.highly_variable_genes(
-        hvg_adata,
+        adata,
     )
     fig = plt.gcf()
     fig.suptitle(f"Highly variable genes dispersion plot - {args.output_prefix}", va="center", ha="center", fontsize=16)

docker/spatial_r/scripts/geomx_counts_to_rds

@@ -85,10 +85,12 @@ geomxdata <- readNanoStringGeoMxSet(dccFiles = dcc_files,
 ## ADD METADATA ##
 ##################
 metadata <- read.csv(args$sample_metadata, header = TRUE, stringsAsFactors = FALSE)
-slide_filename_parts <- strsplit(args$slide_id, "_")[[1]]
-slide_filename_batch <- tail(slide_filename_parts, 1)
+slide_filename_batch <- sub(".*BATCH_", "", args$slide_id)
 # Per slide by batch/run
 filtered_metadata <- metadata[metadata$batch == slide_filename_batch, ]
+if (nrow(filtered_metadata) == 0) {
+	stop("No metadata found for batch: ", slide_filename_batch)
+}
 filtered_metadata$sample <- paste0(filtered_metadata$ASAP_sample_id, "_", filtered_metadata$replicate)
 filtered_metadata <- filtered_metadata[, c("sample_id", "batch", "sample")]
 # Some batches contain multiple replicates for the same ROI which will be in one DCC file, need to group
@@ -107,8 +109,8 @@ merged_pdata <- merge(
 	sort = FALSE
 )
 merged_pdata$batch_id <- paste0(args$team_id, "_", args$dataset_id, "_", merged_pdata$batch)
-num_pdata_rows = length(rownames(pData(geomxdata)))
-num_metadata_rows = length(rownames(merged_pdata))
+num_pdata_rows = nrow(pData(geomxdata))
+num_metadata_rows = nrow(merged_pdata)
 if (num_pdata_rows != num_metadata_rows) {
 	message(glue(
 		"DCC files AKA RDS object row names:\n",