identify and resolve heteroatom issue in distance network

Author: jcrozum

procaliper/_protein.py

@@ -299,7 +299,13 @@ def get_biopython_residues(self) -> list[Residue]:
             raise ValueError("PDB location not set; use `fetch_pdb` first")
         p = PDBParser(QUIET=True)
         structure = p.get_structure("", self.pdb_location_absolute)
-        reslist = [res for model in structure for chain in model for res in chain]
+        reslist = [
+            res
+            for model in structure
+            for chain in model
+            for res in chain
+            if res.get_id()[0] == " "  # excludes heteroatoms and water
+        ]
         return reslist
 
     def get_confidence(self) -> list[float]:

procaliper/network.py

@@ -96,9 +96,7 @@ def regulatory_distance_network(protein: Protein) -> nx.Graph:
     all_regs = {**ptms, **binding, **active, **regions, **domains}
 
     # residues, excluding heteroatoms and water
-    protein_residues = [
-        res for res in protein.get_biopython_residues() if res.get_id()[0] == " "
-    ]
+    protein_residues = protein.get_biopython_residues()
 
     all_regs_residues = {}
     for k, v in all_regs.items():

procaliper/protein_structure/distance.py

@@ -91,7 +91,13 @@ def distance_matrix(
         npt.NDArray[np.float64]: distance matrix with shape nxn where n is the
             number of residues in the structure.
     """
-    residues = [res for model in structure for chain in model for res in chain]
+    residues = [
+        res
+        for model in structure
+        for chain in model
+        for res in chain
+        if res.get_id()[0] == " "
+    ]
     residues = list(enumerate(residues))
     adj = np.ones((len(residues), len(residues))) * np.inf
 

tests/test_network.py

@@ -1,5 +1,5 @@
 from procaliper import Protein
-from procaliper.network import regulatory_distance_network
+from procaliper.network import distance_network, regulatory_distance_network
 
 INCOMPLETE_PDB_PATH = "tests/test_data/1nhz.pdb"
 INCOMPLETE_PDB_UNIPROT_ID = "P04150"
@@ -31,5 +31,5 @@ def test_pdb_with_ligand() -> None:
     oxu = Protein.from_uniprot_id(LIGAND_PDB_UNIPROT_ID)
     oxu.register_local_pdb(path_to_pdb_file=LIGAND_PDB_PATH)
 
-    reg = regulatory_distance_network(oxu)
+    reg = distance_network(oxu)
     assert reg is not None