Various memory and method fixes (#53)

Author: LouisK92

src/methods_cell_type_annotation/moscot/config.vsh.yaml

@@ -57,4 +57,4 @@ runners:
   - type: executable
   - type: nextflow
     directives:
-      label: [ midtime, midcpu, midmem ]
+      label: [ hightime, midcpu, veryhighmem ]

src/methods_cell_type_annotation/ssam/config.vsh.yaml

@@ -35,4 +35,4 @@ runners:
   - type: executable
   - type: nextflow
     directives:
-      label: [ midtime, midcpu, midmem ]
+      label: [ hightime, midcpu, midmem ]

src/methods_expression_correction/resolvi_correction/config.vsh.yaml

@@ -54,4 +54,4 @@ runners:
   - type: executable
   - type: nextflow
     directives:
-      label: [ midtime, highcpu, highmem ]
+      label: [ hightime, highcpu, highmem, gpu ]

src/methods_normalization/spanorm/config.vsh.yaml

@@ -36,4 +36,4 @@ runners:
   - type: executable
   - type: nextflow
     directives:
-      label: [ midtime, lowcpu, lowmem ]
+      label: [ veryhightime, midcpu, highmem ]

src/methods_normalization/spanorm/script.R

@@ -26,7 +26,17 @@ spatial_coords <- matrix(c(centroid_x, centroid_y), ncol = 2)
 spatialCoords(sce) <- spatial_coords
 
 # Apply SpaNorm normalization to the spatial data
-result <- SpaNorm(sce)
+set.seed(777)
+
+# Calculate sample.p based on number of cells (following recommendations of https://github.com/bhuvad/SpaNorm/issues/10#issuecomment-2918248535)
+n_cells <- ncol(sce)
+if (n_cells > 100000) {
+  sample_p <- 25000 / n_cells
+} else {
+  sample_p <- 0.25
+}
+
+result <- SpaNorm(sce, sample.p = sample_p)
 
 # Get the normalized matrix from SpaNorm result (log-transformed normalized counts)
 normalized_matrix <- assay(result, "logcounts")

src/methods_segmentation/binning/config.vsh.yaml

@@ -33,4 +33,4 @@ runners:
   - type: executable
   - type: nextflow
     directives:
-      label: [ midtime, midcpu, midmem ]
+      label: [ midtime, midcpu, highmem ]

src/methods_segmentation/cellpose/config.vsh.yaml

@@ -93,4 +93,4 @@ runners:
   - type: executable
   - type: nextflow
     directives:
-      label: [ midtime, midcpu, highmem, gpu ]
+      label: [ midtime, midcpu, veryhighmem, gpu ]

src/methods_transcript_assignment/basic_transcript_assignment/config.vsh.yaml

@@ -35,4 +35,4 @@ runners:
   - type: executable
   - type: nextflow
     directives:
-      label: [ midtime, lowcpu, lowmem ]
+      label: [ midtime, midcpu, midmem ]

src/methods_transcript_assignment/baysor/config.vsh.yaml

@@ -110,4 +110,4 @@ runners:
   - type: executable
   - type: nextflow
     directives:
-      label: [ midtime, midcpu, midmem ]
+      label: [ veryhightime, midcpu, highmem ]

src/methods_transcript_assignment/baysor/script.py

@@ -29,7 +29,8 @@
 }
 meta = {
   'name': 'baysor_transcript_assignment',
-  'temp_dir': "./temp/methods/baysor"
+  'temp_dir': "./temp/methods/baysor",
+  'cpus': 4,
 }
 ## VIASH END
 
@@ -96,6 +97,9 @@
 
 
 # Run Baysor
+n_threads = meta['cpus'] or os.cpu_count()
+n_threads = max(n_threads-2, 1)
+os.environ['JULIA_NUM_THREADS'] = str(n_threads)
 print('Running Baysor', flush=True)
 baysor_cmd = f"baysor run -c {CONFIG_TOML} -o {BAYSOR_OUTPUT} {TRANSCRIPTS_CSV} {SEGMENTATION_TIF}"
 print("\t" + baysor_cmd, flush=True)