[pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

Author: pre-commit-ci[bot]

tmp_scripts/compare_helpers.py

@@ -1,11 +1,16 @@
+from __future__ import annotations
+
 import warnings
+
 warnings.filterwarnings("ignore")
-import cupy as cp
-import numpy as np
-import anndata as ad
-from pathlib import Path
 import os
 import time
+from pathlib import Path
+
+import anndata as ad
+import cupy as cp
+import numpy as np
+
 import rapids_singlecell as rsc
 
 # Add utils to path for GPU version
@@ -18,64 +23,71 @@ def compare_indices(adata_cpu):
     Compare the saturated/unsaturated indices between CPU and GPU versions.
     """
     print("🔍 Comparing CPU vs GPU index computation...")
-    
+
     # Import the helper functions
     from utils.sepal_cpu import _compute_idxs
     from utils.sepal_gpu import _compute_idxs_gpu
-    
+
     # Get connectivity and spatial data
     adata_gpu = ad.read_h5ad(HOME / "data/visium_hne_adata.h5ad")
     rsc.get.anndata_to_GPU(adata_gpu, convert_all=True)
-    g = adata_cpu.obsp['spatial_connectivities']
-    g_gpu = adata_gpu.obsp['spatial_connectivities']
+    g = adata_cpu.obsp["spatial_connectivities"]
+    g_gpu = adata_gpu.obsp["spatial_connectivities"]
     degrees = cp.diff(g_gpu.indptr)
-    spatial_cpu = adata_cpu.obsm['spatial'].astype(np.float64)
-    spatial_gpu = adata_gpu.obsm['spatial'].astype(cp.float32)
-    
+    spatial_cpu = adata_cpu.obsm["spatial"].astype(np.float64)
+    spatial_gpu = adata_gpu.obsm["spatial"].astype(cp.float32)
+
     g_gpu = rsc.get.X_to_GPU(g_gpu)
     spatial_gpu = rsc.get.X_to_GPU(spatial_gpu)
 
     # Compute indices with both methods
     start = time.time()
     degrees = cp.diff(g_gpu.indptr)
-    sat_gpu, sat_idx_gpu, unsat_gpu, unsat_idx_gpu = _compute_idxs_gpu(g_gpu, degrees, spatial_gpu, 6)
+    sat_gpu, sat_idx_gpu, unsat_gpu, unsat_idx_gpu = _compute_idxs_gpu(
+        g_gpu, degrees, spatial_gpu, 6
+    )
     end = time.time()
     print("GPU indices computed in ", end - start, "seconds")
-    
+
     start = time.time()
-    sat_cpu, sat_idx_cpu, unsat_cpu, unsat_idx_cpu = _compute_idxs(g, spatial_cpu, 6, "l1")
+    sat_cpu, sat_idx_cpu, unsat_cpu, unsat_idx_cpu = _compute_idxs(
+        g, spatial_cpu, 6, "l1"
+    )
     end = time.time()
     print("CPU indices computed in ", end - start, "seconds")
     # Convert GPU results to CPU for comparison
     sat_gpu_cpu = sat_gpu.get()
     sat_idx_gpu_cpu = sat_idx_gpu.get()
     unsat_gpu_cpu = unsat_gpu.get()
     unsat_idx_gpu_cpu = unsat_idx_gpu.get()
-    
+
     print(f"Saturated nodes - CPU: {len(sat_cpu)}, GPU: {len(sat_gpu_cpu)}")
     print(f"Saturated nodes identical: {np.array_equal(sat_cpu, sat_gpu_cpu)}")
-    
-    print(f"Unsaturated nodes - CPU: {len(unsat_cpu)}, GPU: {len(unsat_gpu_cpu)}")  
+
+    print(f"Unsaturated nodes - CPU: {len(unsat_cpu)}, GPU: {len(unsat_gpu_cpu)}")
     print(f"Unsaturated nodes identical: {np.array_equal(unsat_cpu, unsat_gpu_cpu)}")
-    
-    print(f"Saturated indices identical: {np.array_equal(sat_idx_cpu, sat_idx_gpu_cpu)}")
-    
+
+    print(
+        f"Saturated indices identical: {np.array_equal(sat_idx_cpu, sat_idx_gpu_cpu)}"
+    )
+
     # Check unsat_idx differences (these might differ due to tie-breaking)
     unsat_idx_diff = np.sum(unsat_idx_cpu != unsat_idx_gpu_cpu)
-    print(f"Unsaturated index differences: {unsat_idx_diff}/{len(unsat_idx_cpu)} ({100*unsat_idx_diff/len(unsat_idx_cpu):.1f}%)")
-    
+    print(
+        f"Unsaturated index differences: {unsat_idx_diff}/{len(unsat_idx_cpu)} ({100 * unsat_idx_diff / len(unsat_idx_cpu):.1f}%)"
+    )
+
     return {
-        'sat_identical': np.array_equal(sat_cpu, sat_gpu_cpu),
-        'unsat_identical': np.array_equal(unsat_cpu, unsat_gpu_cpu),
-        'sat_idx_identical': np.array_equal(sat_idx_cpu, sat_idx_gpu_cpu),
-        'unsat_idx_diff_count': unsat_idx_diff,
-        'unsat_idx_diff_percent': 100*unsat_idx_diff/len(unsat_idx_cpu)
+        "sat_identical": np.array_equal(sat_cpu, sat_gpu_cpu),
+        "unsat_identical": np.array_equal(unsat_cpu, unsat_gpu_cpu),
+        "sat_idx_identical": np.array_equal(sat_idx_cpu, sat_idx_gpu_cpu),
+        "unsat_idx_diff_count": unsat_idx_diff,
+        "unsat_idx_diff_percent": 100 * unsat_idx_diff / len(unsat_idx_cpu),
     }
 
+
 if __name__ == "__main__":
     # Run comparison
     adata_cpu = ad.read_h5ad(HOME / "data/visium_hne_adata.h5ad")
     res = compare_indices(adata_cpu)
     print(res)
-
-    

tmp_scripts/debug_comparison.py

@@ -1,40 +1,50 @@
+from __future__ import annotations
+
 import warnings
+
 warnings.filterwarnings("ignore")
-import numpy as np
-import cupy as cp
+import os
+from pathlib import Path
+
+import anndata as ad
 from utils.sepal_cpu import sepal
 from utils.sepal_gpu import sepal_gpu
-import anndata as ad
-from pathlib import Path
-import os
+
 import rapids_singlecell as rsc
 
 HOME = Path(os.path.expanduser("~"))
 
+
 def debug_sepal_comparison():
     # Load data
     adata = ad.read_h5ad(HOME / "data/visium_hne_adata.h5ad")
-    
+
     # Test with single gene first
-    test_genes = [ "Npbwr1"]  # Just one gene
-    
+    test_genes = ["Npbwr1"]  # Just one gene
+
     print("=== CPU vs GPU Comparison ===")
-    
+
     # CPU version
-    cpu_result = sepal(adata, max_neighs=6, genes=test_genes, n_iter=4000, copy=True, debug=True)
+    cpu_result = sepal(
+        adata, max_neighs=6, genes=test_genes, n_iter=4000, copy=True, debug=True
+    )
     print(f"CPU result: {cpu_result.iloc[0, 0]}")
-    
-    # GPU version  
+
+    # GPU version
     adata = ad.read_h5ad(HOME / "data/visium_hne_adata.h5ad")
     rsc.get.anndata_to_GPU(adata, convert_all=True)
-    adata.obsp['spatial_connectivities'] = rsc.get.X_to_GPU(adata.obsp['spatial_connectivities'])
-    adata.obsm['spatial'] = rsc.get.X_to_GPU(adata.obsm['spatial'])
-    
+    adata.obsp["spatial_connectivities"] = rsc.get.X_to_GPU(
+        adata.obsp["spatial_connectivities"]
+    )
+    adata.obsm["spatial"] = rsc.get.X_to_GPU(adata.obsm["spatial"])
 
-    gpu_result = sepal_gpu(adata, max_neighs=6, genes=test_genes, n_iter=4000, copy=True, debug=True)
+    gpu_result = sepal_gpu(
+        adata, max_neighs=6, genes=test_genes, n_iter=4000, copy=True, debug=True
+    )
     print(f"GPU result: {gpu_result.iloc[0, 0]}")
-    
+
     print(f"Difference: {abs(cpu_result.iloc[0, 0] - gpu_result.iloc[0, 0])}")
 
+
 if __name__ == "__main__":
     debug_sepal_comparison()

tmp_scripts/prepare_data.py

@@ -1,10 +1,13 @@
-import squidpy as sq
+from __future__ import annotations
 
-from pathlib import Path
 import os
+from pathlib import Path
+
+import squidpy as sq
+
 HOME = Path(os.path.expanduser("~"))
 if __name__ == "__main__":
     (HOME / "data").mkdir(parents=True, exist_ok=True)
     adata = sq.datasets.visium_hne_adata()
     sq.gr.spatial_neighbors(adata)
-    adata.write_h5ad(HOME / "data/visium_hne_adata.h5ad")
+    adata.write_h5ad(HOME / "data/visium_hne_adata.h5ad")

tmp_scripts/run_cpu.py

@@ -1,13 +1,15 @@
+from __future__ import annotations
+
 import warnings
+
 warnings.filterwarnings("ignore")
-import squidpy as sq
-import time
-import anndata as ad
-from pathlib import Path
 import os
+import time
 from argparse import ArgumentParser
+from pathlib import Path
+
+import anndata as ad
 from utils.sepal_cpu import sepal
-import scanpy as sc
 
 HOME = Path(os.path.expanduser("~"))
 
@@ -20,8 +22,10 @@
     genes = ["Gm29570"]
     genes = adata.var_names.values[:10]
     # sc.pp.normalize_total(adata)
-    result = sepal(adata, max_neighs=6, genes=genes, n_iter=30000, copy=True, debug=args.debug)
+    result = sepal(
+        adata, max_neighs=6, genes=genes, n_iter=30000, copy=True, debug=args.debug
+    )
     end_time = time.time()
     print(f"Time taken: {end_time - start_time} seconds")
     result.sort_values(by="sepal_score", ascending=False, inplace=True)
-    print(result.head(10))
+    print(result.head(10))

tmp_scripts/run_gpu.py

@@ -1,16 +1,18 @@
+from __future__ import annotations
+
 import warnings
+
 warnings.filterwarnings("ignore")
 
-import squidpy as sq
+import os
 import time
-import anndata as ad
+from argparse import ArgumentParser
 from pathlib import Path
-import os
-import rapids_singlecell as rsc
-import scanpy as sc
+
+import anndata as ad
 from utils.sepal_gpu import sepal_gpu
-from argparse import ArgumentParser
 
+import rapids_singlecell as rsc
 
 HOME = Path(os.path.expanduser("~"))
 if __name__ == "__main__":
@@ -20,14 +22,18 @@
     adata = ad.read_h5ad(HOME / "data/visium_hne_adata.h5ad")
     # sc.pp.normalize_total(adata)
     rsc.get.anndata_to_GPU(adata, convert_all=True)
-    adata.obsp['spatial_connectivities'] = rsc.get.X_to_GPU(adata.obsp['spatial_connectivities'])
-    adata.obsm['spatial'] = rsc.get.X_to_GPU(adata.obsm['spatial'])
+    adata.obsp["spatial_connectivities"] = rsc.get.X_to_GPU(
+        adata.obsp["spatial_connectivities"]
+    )
+    adata.obsm["spatial"] = rsc.get.X_to_GPU(adata.obsm["spatial"])
     start_time = time.time()
     genes = adata.var_names.values[:1000]
     # genes = ["Gm29570"]
-    result = sepal_gpu(adata, max_neighs=6, genes=genes, n_iter=30000, copy=True, debug=args.debug)
+    result = sepal_gpu(
+        adata, max_neighs=6, genes=genes, n_iter=30000, copy=True, debug=args.debug
+    )
     end_time = time.time()
     print(f"Time taken: {end_time - start_time} seconds")
-    
+
     result.sort_values(by="sepal_score", ascending=False, inplace=True)
-    print(result.head(10))
+    print(result.head(10))