Clustering small detected pockets

.gitignore

@@ -11,6 +11,7 @@ public
 
 __pycache__
 .vscode
+build
 
 # Coverage
 .coverage

README.md

@@ -1,6 +1,6 @@
 <h1 align="center">POVME</h1>
 
-<h4 align="center">Detect and characterize pockets from molecular simulations.</h4>
+<h4 align="center">Detect and characterize protein pockets.</h4>
 
 <p align="center">
     <a href="https://github.com/durrantlab/POVME/actions/workflows/tests.yml">
@@ -17,8 +17,8 @@
     </a>
 </p>
 
-POVME (POcket Volume MEasurer) is an open-source tool designed for detailed analysis of ligand-binding pocket shapes and volumes in proteins.
-Initially developed to support structure-guided drug discovery, POVME has been widely adopted for its simplicity, speed, and flexibility.
+POVME (**PO**cket **V**olume **ME**asurer) is an open-source tool designed for detailed analysis of ligand-binding pocket shapes and volumes in proteins.
+Initially developed to support structure-based drug discovery, POVME has been widely adopted for its simplicity, speed, and flexibility.
 
 ## Why binding pocket analysis?
 
@@ -42,7 +42,7 @@ pixi install
 Now you can activate the new virtual environment using
 
 ```sh
-pixi shell
+pixi shell -e dev
 ```
 
 ## Installation
@@ -95,8 +95,8 @@ This will install the specified version of `povme` and its dependencies into you
 
 If you use POVME in your work, please cite:
 
-1.  Durrant, J.D., L. Votapka, J. Sørensen, and R. E. Amaro (2014). "POVME 2.0: An Enhanced Tool for Determining Pocket Shape and Volume Characteristics." J. Chem. Theory Comput. 10(11):5047-5056.
-2.  Durrant, J. D., C. A. de Oliveira, et al. (2011). "POVME: An algorithm for measuring binding-pocket volumes." J Mol Graph Model 29(5): 773-776.
+1.  Durrant, J. D., Votapka, L., Sørensen, J., & Amaro, R. E. (2014). POVME 2.0: An Enhanced Tool for Determining Pocket Shape and Volume Characteristics. *J. Chem. Theory Comput. 10*(11), 5047-5056.
+2.  Durrant, J. D., de Oliveira, C. A. F., & McCammon, J. A. (2011). POVME: An algorithm for measuring binding-pocket volumes. *J. Mol. Graph. Model. 29*(5), 773-776.
 
 ## License
 

pixi.toml

@@ -9,7 +9,7 @@ readme = "README.md"
 
 [pypi-dependencies]
 povme = { path = ".", editable = true }
-pymolecule = { git = "https://github.com/durrantlab/pymolecule.git", rev = "121c34103169d362f4ed2ca9207dd5555f9dc25b" }
+pymolecule = { git = "https://github.com/durrantlab/pymolecule.git", rev = "423cf64f17c21881a1685402437bd69a81e79ccf" }
 ray = { version = ">=2.37.0,<3"}
 
 [system-requirements]

povme/pocket/detect.py

@@ -3,7 +3,7 @@
 import numpy as np
 from loguru import logger
 from pymolecule import Molecule
-from scipy.cluster.vq import kmeans2
+from scipy.cluster.vq import ClusterError, kmeans2
 from scipy.spatial.distance import cdist
 
 from povme.config import PocketDetectConfig
@@ -29,7 +29,6 @@ def __init__(
         self.config = PocketDetectConfig()
 
     def run(self, path_pdb: str, output_prefix: str = "") -> None:
-
         config = self.config
 
         # If the output prefix includes a directory, create that directory if
@@ -185,11 +184,32 @@ def run(self, path_pdb: str, output_prefix: str = "") -> None:
 
             # do I need to whiten stuff here? not sure what whitening is.
 
-            centroids, idx = kmeans2(pts, config.n_spheres)
+            n_points = pts.shape[0]
+            logger.debug("Pocket has {} points", n_points)
+            n_clusters = min(config.n_spheres, n_points)
+            try:
+                _, idx = kmeans2(pts, n_clusters, missing="raise")
+            except ClusterError:
+                logger.warning("One of the kmeans clusters are empty")
+                logger.warning("Retrying kmeans with different initialization")
+                try:
+                    _, idx = kmeans2(pts, n_clusters, minit="++", missing="raise")
+                except ClusterError:
+                    logger.warning("kmeans still has empty clusters")
+                    logger.warning("We are using all points instead")
+                    idx = np.arange(0, n_points)
+                except Exception as e:
+                    raise e
+            except Exception as e:
+                raise e
+            finally:
+                cluster_labels = np.unique(idx)
+            logger.debug(f"Found clusters of {cluster_labels}")
 
             pts_string = ""
-            for cluster_num in range(config.n_spheres):
+            for cluster_num in cluster_labels:
                 indexes_for_this_cluster = np.nonzero(idx == cluster_num)[0]
+                logger.debug(f"Point indices for cluster: {indexes_for_this_cluster}")
                 cluster_pts = pts[indexes_for_this_cluster]
                 cluster_center = np.mean(cluster_pts, axis=0)
                 try:

pyproject.toml

@@ -1,13 +1,15 @@
 [project]
+name = "povme"
+version = "2.2.1"
+description = "Detect and characterize protein pockets."
 authors = [
     {name = "Durrant Lab", email = "durrantj@pitt.edu"}
 ]
 maintainers = [
-    {name = "Jacob Durrant", email = "durrantj@pitt.edu"}
+    {name = "Jacob Durrant", email = "durrantj@pitt.edu"},
+    {name = "Alex Maldonado", email = "alex.maldonado@pitt.edu"},
 ]
-description = "TODO"
-name = "povme"
-version = "2.2.1"
+license = "Apache-2.0"
 readme = "README.md"
 requires-python = ">=3.10,<3.13"
 # TODO: Keep this here until pixi releases building capabilities
@@ -18,13 +20,20 @@ dependencies = [
     "pyyaml>=6.0.2,<7",
     "loguru>=0.7.2,<0.8",
     "ray>=2.38.0,<3",
+    "pymolecule@git+https://github.com/durrantlab/pymolecule",
 ]
 
+[project.urls]
+Repository = "https://github.com/durrantlab/POVME"
+Documentation = "https://durrantlab.github.io/POVME/"
+Homepage = "https://durrantlab.github.io/POVME/"
+
 [project.scripts]
 povme = "povme.cli:povme_cli"
 
 [build-system]
-requires = ["setuptools"]
+requires = ["setuptools", "wheel"]
+build-backend = "setuptools.build_meta"
 
 [tool.setuptools.packages.find]
 where = ["."]

tests/conftest.py

@@ -54,6 +54,16 @@ def path_rel1_output():
     return os.path.join(TEST_DIR, "tmp/", "rel1/")
 
 
+@pytest.fixture
+def path_5csn_config():
+    return os.path.join(TEST_DIR, "files/5csn/pocket-id.yml")
+
+
+@pytest.fixture
+def path_5csn_output():
+    return os.path.join(TEST_DIR, "tmp/", "5csn/")
+
+
 @pytest.fixture
 def path_rogfp2_pdb():
     return os.path.join(TEST_DIR, "files/rogfp2/rogfp2.pdb")