add singler python implementation (#118)

* add singler python implementation

* script to run the benchmark

* workflow file

* workflow configuration

* improvements

* Adjust docs and repo in singler

---------

Co-authored-by: LouisK92 <[email protected]>

Author: dariarom94

scripts/run_benchmark/run_full_local.sh

@@ -59,6 +59,7 @@ celltype_annotation_methods:
   # - moscot
   # - mapmycells
   # - tangram
+  # - singler
 expression_correction_methods:
   - no_correction
   # - gene_efficiency_correction

scripts/run_benchmark/run_full_seqeracloud.sh

@@ -51,6 +51,7 @@ celltype_annotation_methods:
   - moscot
   - mapmycells
   - tangram
+  - singler
 expression_correction_methods:
   - no_correction
   - gene_efficiency_correction

scripts/run_benchmark/run_test_local.sh

@@ -54,6 +54,7 @@ celltype_annotation_methods:
   # - moscot
   # - mapmycells
   # - tangram
+  # - singler
 expression_correction_methods:
   - no_correction
   # - gene_efficiency_correction

scripts/run_benchmark/run_test_seqeracloud.sh

@@ -50,6 +50,7 @@ celltype_annotation_methods:
   - moscot
   - mapmycells
   - tangram
+  - singler
 expression_correction_methods:
   - no_correction
   - gene_efficiency_correction

src/methods_cell_type_annotation/singler/config.vsh.yaml

@@ -0,0 +1,38 @@
+__merge__: /src/api/comp_method_cell_type_annotation.yaml
+
+name: singler
+label: "singler"
+summary: "Cell type annotations using single-cell reference with SingleR"
+description: "Cell type annotations using single-cell reference with SingleR"
+
+links:
+  documentation: "https://github.com/SingleR-inc/singler-py"
+  repository: "https://github.com/SingleR-inc/singler-py"
+references:
+  doi: "10.1038/s41590-018-0276-y"
+
+arguments:
+  - name: --labels_key
+    type: string
+    description: The key of the cell labels in the input data.
+    default: cell_labels
+
+resources:
+  - type: python_script
+    path: script.py
+
+engines:
+  - type: docker
+    image: openproblems/base_python:1
+    setup:
+    - type: python
+      pypi: [singler]
+    __merge__: 
+      - /src/base/setup_spatialdata_partial.yaml
+  - type: native
+
+runners:
+  - type: executable
+  - type: nextflow
+    directives:
+      label: [ midtime, midcpu, midmem ]

src/methods_cell_type_annotation/singler/script.py

@@ -0,0 +1,53 @@
+import anndata as ad
+import os
+import shutil
+
+import singlecellexperiment as sce
+import singler
+
+## VIASH START
+# The following code has been auto-generated by Viash.
+par = {
+  'input_spatial_normalized_counts': r'resources_test/task_ist_preprocessing/mouse_brain_combined/spatial_normalized_counts.h5ad',
+  'input_transcript_assignments': r'resources_test/task_ist_preprocessing/mouse_brain_combined/transcript_assignments.zarr',
+  'input_scrnaseq_reference': r'resources_test/task_ist_preprocessing/mouse_brain_combined/scrnaseq_reference.h5ad',
+  'celltype_key': r'cell_type',
+  'output': r'resources_test/task_ist_preprocessing/mouse_brain_combined/spatial_with_cell_types.h5ad',
+  'labels_key': r'cell_labels'
+}
+meta = {
+  'name': r'singleR',
+  'functionality_name': r'singleR'
+}
+dep = {
+  
+}
+
+## VIASH END
+sce_h5ad = sce.read_h5ad(par['input_spatial_normalized_counts'])
+adata_sp = ad.read_h5ad(par['input_spatial_normalized_counts'])
+
+sce_ref = sce.read_h5ad(par['input_scrnaseq_reference'])
+
+features = [str(x) for x in sce_h5ad.row_data.row_names]
+
+mat = sce_h5ad.assay("counts") ##example has raw, not sure
+mat = mat.sorted_indices() ## magic line to make sure the matrix is in the right format for SingleR
+
+mat_ref = sce_ref.assay("normalized")
+mat_ref = mat_ref.sorted_indices() ## magic line to make sure the matrix is in the right format for SingleR
+
+## create the reference from our sc data
+built = singler.train_single(ref_data = mat_ref,   
+                             ref_labels = sce_ref.get_column_data().column("cell_type"),    
+                             ref_features = sce_ref.get_row_names(),    
+                             test_features = features,)
+
+## annotate the dataset
+output = singler.classify_single(mat, ref_prebuilt=built)
+
+adata_sp.obs["cell_type"] = output['best']
+
+# Write output
+print('Writing output', flush=True)
+adata_sp.write(par['output'])

src/workflows/run_benchmark/config.vsh.yaml

@@ -98,7 +98,7 @@ argument_groups:
           A list of cell type annotation methods to run.
         type: string
         multiple: true
-        default: "ssam:tacco:moscot:mapmycells:tangram"
+        default: "ssam:tacco:moscot:mapmycells:tangram:singler"
       - name: "--expression_correction_methods"
         description: |
           A list of expression correction methods to run.
@@ -170,6 +170,7 @@ dependencies:
   - name: methods_cell_type_annotation/moscot
   - name: methods_cell_type_annotation/mapmycells
   - name: methods_cell_type_annotation/tangram
+  - name: methods_cell_type_annotation/singler
   - name: methods_expression_correction/no_correction
   - name: methods_expression_correction/gene_efficiency_correction
   - name: methods_expression_correction/resolvi_correction

src/workflows/run_benchmark/main.nf

@@ -376,7 +376,8 @@ workflow run_wf {
     tacco,
     moscot,
     mapmycells,
-    tangram
+    tangram,
+    singler
   ]
 
   cta_ch = normalization_ch