Merge pull request #7 from HiDiHlabs/dev

Prepare new release

Author: niklasmueboe

.pre-commit-config.yaml

@@ -15,7 +15,7 @@ repos:
       - id: no-commit-to-branch
         args: [--branch=main]
   - repo: https://github.com/astral-sh/ruff-pre-commit
-    rev: v0.6.9
+    rev: v0.7.4
     hooks:
       - id: ruff
         args: [--fix]
@@ -28,7 +28,7 @@ repos:
     hooks:
       - id: black
   - repo: https://github.com/pre-commit/mirrors-mypy
-    rev: v1.11.2
+    rev: v1.13.0
     hooks:
       - id: mypy
   - repo: https://github.com/codespell-project/codespell

README.md

@@ -31,7 +31,7 @@ conda install bioconda::spatialleiden
 ```
 
 For detailed installation instructions please refer to the
-[documentation](https://spatialleiden.readthedocs.io/stable/installation.html).
+[documentation](https://spatialleiden.readthedocs.io/page/installation.html).
 
 ## Documentation
 
@@ -46,6 +46,17 @@ Müller-Bötticher, N., Sahay, S., Eils, R., and Ishaque, N.
 "SpatialLeiden - Spatially-aware Leiden clustering"
 bioRxiv (2024) https://doi.org/10.1101/2024.08.23.609349
 
+```
+@article {spatialleiden2024,
+	author = {Müller-Bötticher, Niklas and Sahay, Shashwat and Eils, Roland and Ishaque, Naveed},
+	title = {SpatialLeiden - Spatially-aware Leiden clustering},
+	year = {2024},
+	doi = {10.1101/2024.08.23.609349},
+	journal = {bioRxiv},
+	publisher = {Cold Spring Harbor Laboratory}
+}
+```
+
 ## Versioning
 
 This project follows the [SemVer](https://semver.org/) guidelines for versioning.

pyproject.toml

@@ -39,9 +39,10 @@ docs = ["sphinx", "sphinx-copybutton", "sphinx-rtd-theme", "squidpy", "myst-nb"]
 dev  = ["spatialleiden[docs]", "pre-commit"]
 
 [project.urls]
-homepage      = "https://github.com/HiDiHlabs/SpatialLeiden"
-documentation = "https://spatialleiden.readthedocs.io"
-repository    = "https://github.com/HiDiHlabs/SpatialLeiden"
+Homepage      = "https://github.com/HiDiHlabs/SpatialLeiden"
+Documentation = "https://spatialleiden.readthedocs.io"
+Repository    = "https://github.com/HiDiHlabs/SpatialLeiden"
+Issues        = "https://github.com/HiDiHlabs/SpatialLeiden/issues"
 
 
 [tool]
@@ -56,12 +57,13 @@ include = ["spatialleiden"]
 profile = "black"
 
 [tool.black]
-target-version = ["py310", "py311", "py312"]
+target-version = ["py310", "py311", "py312", "py313"]
 
 [tool.ruff]
 target-version = "py310"
 
 [tool.mypy]
+python_version         = "3.10"
 ignore_missing_imports = true
 warn_no_return         = false
 packages               = "spatialleiden"

spatialleiden/_multiplex_leiden.py

@@ -13,17 +13,13 @@
 
 
 def _build_igraph(adjacency: _GraphArray, *, directed: bool = True) -> Graph:
-    # adapted from scanpy https://github.com/scverse/scanpy
     sources, targets, weights = find(adjacency)
-    g = Graph(directed=directed)
-    g.add_vertices(adjacency.shape[0])
-    g.add_edges(list(zip(sources, targets)))
-    g.es["weight"] = weights
-    if g.vcount() != adjacency.shape[0]:
-        raise RuntimeError(
-            f"The constructed graph has only {g.vcount()} nodes. "
-            "Your adjacency matrix contained redundant nodes."
-        )
+    g = Graph(
+        n=adjacency.shape[0],
+        edges=zip(sources, targets),
+        directed=directed,
+        edge_attrs={"weight": weights},
+    )
     return g
 
 

spatialleiden/_resolution_search.py

@@ -1,4 +1,5 @@
 from collections.abc import Callable
+from warnings import warn
 
 import scanpy as sc
 from anndata import AnnData
@@ -8,42 +9,38 @@
 
 def _search_resolution(
     fn: Callable[[float], int],
-    ncluster: int,
+    n_clusters: int,
     start: float = 1,
     step: float = 0.1,
     n_iterations: int = 15,
 ) -> float:
-    # adapted from SpaGCN.search_res (https://github.com/jianhuupenn/SpaGCN)
-    res = start
-    old_ncluster = fn(res)
-    iter = 1
-    while old_ncluster != ncluster:
-        old_sign = 1 if (old_ncluster < ncluster) else -1
-        new_ncluster = fn(res + step * old_sign)
-        if new_ncluster == ncluster:
-            res = res + step * old_sign
-            # print(f"Recommended res = {res:.2f}")
-            return res
-        new_sign = 1 if (new_ncluster < ncluster) else -1
-        if new_sign == old_sign:
-            res = res + step * old_sign
-            # print(f"Res changed to {res:.2f}")
-            old_ncluster = new_ncluster
-        else:
-            step = step / 2
-            # print(f"Step changed to {step:.2f}")
-        if iter > n_iterations:
-            # print("Exact resolution not found")
-            # print(f"Recommended res =  {res:.2f}")
-            return res
-        iter += 1
-    # print(f"Recommended res = {res:.2f}")
-    return res
+
+    if n_iterations <= 2:
+        raise ValueError("At least 2 iterations are necessary")
+    resolution = start
+    n = fn(resolution)
+    i = 1
+    while n != n_clusters:
+        if i >= n_iterations:
+            warn(
+                "Correct resolution not found. Consider increasing the number of "
+                "iterations or adjusting the step size."
+            )
+            break
+        sign = 1 if n < n_clusters else -1
+        resolution += step * sign  # TODO what happens when approaching zero
+        n = fn(resolution)
+        new_sign = 1 if n < n_clusters else -1
+        if new_sign != sign:
+            step /= 2
+        i += 1
+
+    return resolution
 
 
 def search_resolution_latent(
     adata: AnnData,
-    ncluster: int,
+    n_clusters: int,
     *,
     start: float = 1,
     step: float = 0.1,
@@ -56,7 +53,7 @@ def search_resolution_latent(
     Parameters
     ----------
     adata : anndata.AnnData
-    ncluster : int
+    n_clusters : int
         Number of clusters.
     start : float, optional
         Starting point for resolution.
@@ -75,17 +72,19 @@ def search_resolution_latent(
     """
 
     def ncluster4res_leiden(resolution: float) -> int:
-        sc.tl.leiden(adata, resolution=resolution, **kwargs)
+        sc.tl.leiden(adata, resolution=resolution, key_added=key_added, **kwargs)
         return adata.obs[key_added].cat.categories.size
 
     key_added = kwargs.pop("key_added", "leiden")
 
-    return _search_resolution(ncluster4res_leiden, ncluster, start, step, n_iterations)
+    return _search_resolution(
+        ncluster4res_leiden, n_clusters, start, step, n_iterations
+    )
 
 
 def search_resolution_spatial(
     adata: AnnData,
-    ncluster: int,
+    n_clusters: int,
     *,
     start: float = 0.4,
     step: float = 0.1,
@@ -102,7 +101,7 @@ def search_resolution_spatial(
     Parameters
     ----------
     adata : anndata.AnnData
-    ncluster : int
+    n_clusters : int
         Number of clusters.
     start : float, optional
         Starting point for resolution.
@@ -133,7 +132,7 @@ def ncluster4res_spatialleiden(resolution: float) -> int:
     resolution_user = kwargs.pop("resolution", (1, 1))
 
     return _search_resolution(
-        ncluster4res_spatialleiden, ncluster, start, step, n_iterations
+        ncluster4res_spatialleiden, n_clusters, start, step, n_iterations
     )