Do not lookup bonds for hetero residues

Author: Emrys-Merlin

src/biotite/structure/bonds.pyx

@@ -1735,6 +1735,7 @@ def connect_via_residue_names(atoms, bint inter_residue=True,
     cdef np.ndarray atom_names = atoms.atom_name
     cdef np.ndarray atom_names_in_res
     cdef np.ndarray res_names = atoms.res_name
+    cdef np.ndarray hetero = atoms.hetero
     cdef str atom_name1, atom_name2
     cdef int64[:] atom_indices1, atom_indices2
     cdef dict bond_dict_for_res
@@ -1745,6 +1746,10 @@ def connect_via_residue_names(atoms, bint inter_residue=True,
         curr_start_i = residue_starts[res_i]
         next_start_i = residue_starts[res_i+1]
 
+        # Skip lookup for ligands
+        if hetero[curr_start_i]:
+            continue
+
         if custom_bond_dict is None:
             bond_dict_for_res = bonds_in_residue(res_names[curr_start_i])
         else:

tests/structure/data/hetatm/ligand.pdb

@@ -0,0 +1,10 @@
+HETATM  704  C7  LIG B 101     -17.432  24.497  -0.918  1.00 26.28           C  
+HETATM  705  C8  LIG B 101     -17.432  24.497  -0.918  1.00 26.28           C  
+HETATM  706  C13 LIG B 101     -17.432  24.497  -0.918  1.00 26.28           C  
+HETATM  707  C14 LIG B 101     -17.432  24.497  -0.918  1.00 26.28           C  
+HETATM  708  C15 LIG B 101     -17.432  24.497  -0.918  1.00 26.28           C  
+CONECT  704  705
+CONECT  705  704  706
+CONECT  706  705  707
+CONECT  707  706 708
+CONECT  708  707

tests/structure/io/test_pdb.py

@@ -468,12 +468,13 @@ def test_bond_parsing():
     pdb_file = pdb.PDBFile.read(path)
     atoms = pdb.get_structure(pdb_file, model=1, include_bonds=True)
 
+    # TODO get input on this modification
+    atoms = atoms[~atoms.hetero]
     test_bonds = atoms.bonds
     test_bonds.remove_bond_order()
 
     ref_bonds = struc.connect_via_residue_names(atoms)
     ref_bonds.remove_bond_order()
-
     assert test_bonds.as_set() == ref_bonds.as_set()
 
 
@@ -572,3 +573,26 @@ def test_setting_incompatible_structure(annotation, value, warning_only):
     else:
         with pytest.raises(struc.BadStructureError):
             pdb_file.set_structure(atoms)
+
+
+def test_hetatm_intra_residue_bonds():
+    """
+    Expect that HETATM intra-residues bonds are only parsed from CONECT records
+    and not looked up via residue names.
+    """
+    expected_bonds = np.array(
+        [
+            [0, 1, 0],
+            [1, 2, 0],
+            [2, 3, 0],
+            [3, 4, 0],
+        ],
+        dtype=np.uint32,
+    )
+    path = join(data_dir("structure"), "hetatm/ligand.pdb")
+
+    pdb_file = pdb.PDBFile.read(path)
+    structure = pdb.get_structure(pdb_file, model=1, include_bonds=True)
+    actual_bonds = structure.bonds.as_array()
+
+    np.testing.assert_array_equal(actual_bonds, expected_bonds)