Add scBSP method

src/methods/scbsp/config.vsh.yaml

@@ -0,0 +1,44 @@
+__merge__: ../../api/comp_method.yaml
+name: scbsp
+label: scBSP
+summary: scBSP identifies spatially variable genes via granularity-based spatial statistics.
+description: |
+  scBSP (single-cell BinSpect) detects spatially variable genes using a
+  granularity-based approach that tests for spatial patterns at multiple
+  resolutions. It bins spatial coordinates into grids at different
+  granularity levels and tests whether gene expression shows significant
+  spatial structure using a permutation-based statistical test. The method
+  is designed for large-scale spatial transcriptomics data.
+references:
+  bibtex:
+    - |
+      @article{li2024scbsp,
+        title={scBSP: A fast and accurate tool for identifying spatially variable
+               genes from spatial transcriptomic data},
+        author={Li, Jinpu and Shang, Yiqing and others},
+        journal={Bioinformatics},
+        year={2024}
+      }
+info:
+  preferred_normalization: counts
+links:
+  repository: https://github.com/juexinwang/scBSP
+resources:
+  - type: python_script
+    path: script.py
+engines:
+  - type: docker
+    image: openproblems/base_python:1.0.0
+    setup:
+      - type: python
+        packages:
+          - pandas
+          - numpy
+          - scipy
+          - anndata
+          - scbsp
+runners:
+  - type: executable
+  - type: nextflow
+    directives:
+      label: [midtime, highmem, lowcpu]

src/methods/scbsp/script.py

@@ -0,0 +1,55 @@
+import warnings
+warnings.filterwarnings('ignore')
+
+import anndata as ad
+import numpy as np
+import pandas as pd
+
+# VIASH START
+par = {
+    'input_data': 'resources_test/task_spatially_variable_genes/mouse_brain_coronal/dataset.h5ad',
+    'output': 'output.h5ad'
+}
+meta = {
+    'name': 'scbsp'
+}
+# VIASH END
+
+print('Load data', flush=True)
+adata = ad.read_h5ad(par['input_data'])
+
+print('Run scBSP', flush=True)
+from scbsp import granp
+
+coords = adata.obsm['spatial']
+X = adata.layers['counts']
+
+if hasattr(X, 'toarray'):
+    X_dense = X.toarray()
+else:
+    X_dense = np.asarray(X)
+
+result_df = granp(
+    np.ascontiguousarray(coords, dtype=np.float64),
+    np.ascontiguousarray(X_dense, dtype=np.float64),
+)
+
+# granp returns DataFrame with columns: gene_names, p_values
+pvals = result_df['p_values'].values
+scores = -np.log10(np.clip(pvals, 1e-300, 1))
+
+df = pd.DataFrame({
+    'feature_id': list(adata.var_names),
+    'pred_spatial_var_score': scores
+})
+
+output = ad.AnnData(
+    var=df,
+    uns={
+        'dataset_id': adata.uns['dataset_id'],
+        'method_id': meta['name']
+    }
+)
+
+print('Write output to file', flush=True)
+output.write_h5ad(par['output'])