helmkit 0.5.0__tar.gz → 0.5.2__tar.gz

{helmkit-0.5.0 → helmkit-0.5.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: helmkit
-Version: 0.5.0
+Version: 0.5.2
 Summary: Parse HELM strings into RDKit molecules
 License-File: LICENSE
 Requires-Python: >=3.11

{helmkit-0.5.0 → helmkit-0.5.2}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "helmkit"
-version = "0.5.0"
+version = "0.5.2"
 description = "Parse HELM strings into RDKit molecules"
 readme = "README.md"
 requires-python = ">=3.11"

{helmkit-0.5.0 → helmkit-0.5.2}/src/helmkit/molecule.py

@@ -1,3 +1,4 @@
+import bisect
 import multiprocessing
 import re
 import warnings
@@ -735,7 +736,42 @@ class Molecule:
 
     def _sanitize(self) -> None:
         """Clean up the molecule by removing dummy atoms."""
-        self.mol = Chem.DeleteSubstructs(self.mol, Chem.MolFromSmarts("[#0]"))
+        pattern = Chem.MolFromSmarts("[#0]")
+        matches = self.mol.GetSubstructMatches(pattern)
+        atoms_to_delete = sorted({idx for match in matches for idx in match})
+        self.mol = Chem.DeleteSubstructs(self.mol, pattern)
+
+        def correction(offset: int, idx: int) -> int:
+            return bisect.bisect_left(atoms_to_delete, idx) - bisect.bisect_left(
+                atoms_to_delete, offset
+            )
+
+        for i, (m1, a1, m2, a2) in enumerate(self.bondlist):
+            offset1 = self.offset[m1]
+            offset2 = self.offset[m2]
+            self.bondlist[i][1] -= correction(offset1, offset1 + a1)
+            self.bondlist[i][3] -= correction(offset2, offset2 + a2)
+
+        self.offset = [
+            offset - sum(d < offset for d in atoms_to_delete) for offset in self.offset
+        ]
+
+    @property
+    def bond_indices(self) -> List[int]:
+        return [
+            self.mol.GetBondBetweenAtoms(
+                self.offset[monomer1_idx] + atom1_idx,
+                self.offset[monomer2_idx] + atom2_idx,
+            ).GetIdx()
+            for monomer1_idx, atom1_idx, monomer2_idx, atom2_idx in self.bondlist
+        ]
+
+    @property
+    def monomer_indices(self) -> List[int]:
+        return [
+            bisect.bisect_right(self.offset, i) - 1
+            for i in range(self.mol.GetNumAtoms())
+        ]
 
 
 def _init_pool(monomer_df: MonomerLibrary):

{helmkit-0.5.0 → helmkit-0.5.2}/uv.lock

@@ -196,7 +196,7 @@ wheels = [
 
 [[package]]
 name = "helmkit"
-version = "0.5.0"
+version = "0.5.2"
 source = { editable = "." }
 dependencies = [
     { name = "rdkit" },