pxmeter 0.1.1__tar.gz → 0.1.3__tar.gz

{pxmeter-0.1.1/pxmeter.egg-info → pxmeter-0.1.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: pxmeter
-Version: 0.1.1
+Version: 0.1.3
 Summary: PXMeter is a comprehensive toolkit for evaluating the quality of         structures generated by biomolecular structure prediction models.
 Author: Bytedance Inc.
 Author-email: ai4s-bio@bytedance.com

{pxmeter-0.1.1 → pxmeter-0.1.3}/pxmeter/calc_metric.py

@@ -34,6 +34,7 @@ from pxmeter.configs.run_config import RUN_CONFIG
 from pxmeter.constants import IONS, LIGAND
 from pxmeter.data.ccd import get_ccd_mol_from_chain_atom_array
 from pxmeter.data.struct import Structure
+from pxmeter.metrics.clashes import check_clashes_by_vdw
 from pxmeter.metrics.lddt_metrics import LDDT
 from pxmeter.metrics.rmsd_metrics import RMSDMetrics
 
@@ -554,6 +555,16 @@ class MetricResult:
         meta_info_dict["ref_to_model_chain_mapping"] = chain_map
         meta_info_dict["ref_chain_info"] = cls._get_chain_info(ref_struct)
 
+        # Calculate clashes
+        if metric_config.calc_clashes:
+            clashes = check_clashes_by_vdw(
+                model_struct.atom_array,
+                vdw_scale_factor=metric_config.clashes.vdw_scale_factor,
+            )
+            complex_result_dict["clashes"] = len(
+                {x for a, b in clashes for x in (a, b)}
+            )
+
         # Calculate RMSD (if ligand and pocket specified in ref_features)
         if metric_config.calc_rmsd and interested_lig_label_asym_id:
             rmsd_metrics = RMSDMetrics(

{pxmeter-0.1.1 → pxmeter-0.1.3}/pxmeter/cli.py

@@ -18,9 +18,10 @@ from pathlib import Path
 import click
 
 from pxmeter.configs.data_config import (
+    CCD_BLOCKS_JSON,
     COMPONENTS_FILE,
     download_ccd_cif,
-    make_one_letter_code_json_from_ccd,
+    make_precomputed_json_from_ccd,
     ONE_LETTER_CODE_JSON,
 )
 from pxmeter.eval import evaluate, MetricResult
@@ -176,4 +177,6 @@ def update():
     Update the CCD database.
     """
     download_ccd_cif(output_path=COMPONENTS_FILE.parent)
-    make_one_letter_code_json_from_ccd(COMPONENTS_FILE, ONE_LETTER_CODE_JSON)
+    make_precomputed_json_from_ccd(
+        COMPONENTS_FILE, CCD_BLOCKS_JSON, ONE_LETTER_CODE_JSON
+    )

{pxmeter-0.1.1 → pxmeter-0.1.3}/pxmeter/configs/data_config.py

@@ -55,7 +55,9 @@ def download_ccd_cif(output_path: Path):
     logging.info("Download CCD CIF file successfully: %s", output_cif)
 
 
-def make_one_letter_code_json_from_ccd(components_file: Path, output_json: Path):
+def make_precomputed_json_from_ccd(
+    components_file: Path, ccd_blocks_json: Path, one_letter_code_json: Path
+):
     """
     Make a one-letter code JSON file from the CCD CIF file.
 
@@ -65,17 +67,23 @@ def make_one_letter_code_json_from_ccd(components_file: Path, output_json: Path)
     ccd_cif = gemmi.cif.read(str(components_file))
 
     ccd_code_to_one_letter_code = {}
+    ccd_code_to_block = {}
     for block in ccd_cif:
         ccd_code = block.find_value("_chem_comp.id")
         one_letter_code = block.find_value("_chem_comp.one_letter_code")
+        ccd_code_to_block[ccd_code] = block.as_string()
         if one_letter_code is None or one_letter_code == "?":
             continue
         ccd_code_to_one_letter_code[ccd_code] = one_letter_code
 
-    with open(output_json, "w") as f:
-        json.dump(ccd_code_to_one_letter_code, f, indent=4)
+    with open(ccd_blocks_json, "w") as f:
+        json.dump(ccd_code_to_block, f, indent=4)
+
+    logging.info("Make CCD_BLOCKS_JSON successfully: %s", ccd_blocks_json)
 
-    logging.info("Make ONE_LETTER_CODE_JSON successfully: %s", output_json)
+    with open(one_letter_code_json, "w") as f:
+        json.dump(ccd_code_to_one_letter_code, f, indent=4)
+    logging.info("Make ONE_LETTER_CODE_JSON successfully: %s", one_letter_code_json)
 
 
 # default is <repo_dir>/ccd_cache/components.cif Your path for components file
@@ -84,20 +92,30 @@ def make_one_letter_code_json_from_ccd(components_file: Path, output_json: Path)
 repo_dir = Path(__file__).absolute().parent.parent.parent
 ccd_file_in_repo = repo_dir / "ccd_cache" / "components.cif"
 COMPONENTS_FILE = Path(os.environ.get("PXM_CCD_FILE", ccd_file_in_repo))
-ONE_LETTER_CODE_JSON = COMPONENTS_FILE.with_suffix(".json")
+
+CCD_BLOCKS_JSON = COMPONENTS_FILE.with_suffix(".json")
+ONE_LETTER_CODE_JSON = COMPONENTS_FILE.parent / "one_letter_code.json"
 
 if not COMPONENTS_FILE.exists():
     logging.debug(
         "CCD CIF file not found. Downloading CCD CIF file to %s", COMPONENTS_FILE.parent
     )
     download_ccd_cif(output_path=COMPONENTS_FILE.parent)
-    make_one_letter_code_json_from_ccd(COMPONENTS_FILE, ONE_LETTER_CODE_JSON)
+    make_precomputed_json_from_ccd(
+        COMPONENTS_FILE, CCD_BLOCKS_JSON, ONE_LETTER_CODE_JSON
+    )
 else:
     logging.debug("Load CCD CIF file from: %s", COMPONENTS_FILE)
 
-if not ONE_LETTER_CODE_JSON.exists():
-    make_one_letter_code_json_from_ccd(COMPONENTS_FILE, ONE_LETTER_CODE_JSON)
+if not ONE_LETTER_CODE_JSON.exists() or not CCD_BLOCKS_JSON.exists():
+    make_precomputed_json_from_ccd(
+        COMPONENTS_FILE, CCD_BLOCKS_JSON, ONE_LETTER_CODE_JSON
+    )
 
 logging.debug("Load CCD one-letter code from: %s", ONE_LETTER_CODE_JSON)
 with open(ONE_LETTER_CODE_JSON, "r") as f:
     CCD_ONE_LETTER_CODE = json.load(f)
+
+
+with open(CCD_BLOCKS_JSON, "r") as f:
+    CCD_BLOCKS = json.load(f)

{pxmeter-0.1.1 → pxmeter-0.1.3}/pxmeter/configs/run_config.py

@@ -23,6 +23,7 @@ RUN_CONFIG = ConfigDict(
             "enumerate_all_anchors": True,
         },
         "metric": {
+            "calc_clashes": True,
             "calc_lddt": True,
             "calc_dockq": True,
             "calc_rmsd": True,
@@ -32,6 +33,9 @@ RUN_CONFIG = ConfigDict(
                 "nucleotide_threshold": 30.0,
                 "non_nucleotide_threshold": 15.0,
             },
+            "clashes": {
+                "vdw_scale_factor": 0.5,
+            },
         },
     }
 )

{pxmeter-0.1.1 → pxmeter-0.1.3}/pxmeter/data/ccd.py

@@ -24,7 +24,7 @@ from pdbeccdutils.core import ccd_reader
 from rdkit import Chem
 from rdkit.Geometry import Point3D
 
-from pxmeter.configs.data_config import COMPONENTS_FILE
+from pxmeter.configs.data_config import CCD_BLOCKS
 from pxmeter.data.utils import (
     get_inter_residue_bonds,
     get_mol_graph_matches,
@@ -34,17 +34,6 @@ from pxmeter.data.utils import (
 logging.getLogger("rdkit").setLevel(logging.ERROR)
 
 
-@functools.lru_cache
-def _gemmi_load_ccd_cif() -> gemmi.cif.Document:
-    """
-    Load the CCD components file using the gemmi library.
-
-    Returns:
-        gemmi.cif.Document: gemmi ccd components file
-    """
-    return gemmi.cif.read(str(COMPONENTS_FILE))
-
-
 @functools.lru_cache
 def get_ccd_mol_from_cif(ccd_code: str) -> Chem.Mol:
     """
@@ -57,10 +46,8 @@ def get_ccd_mol_from_cif(ccd_code: str) -> Chem.Mol:
         mol (Chem.Mol): The RDKit molecule object corresponding to the given CCD code.
                               Returns None if the CCD code is not found.
     """
-    ccd_cif = _gemmi_load_ccd_cif()
-
     try:
-        ccd_block = ccd_cif[ccd_code]
+        ccd_block = gemmi.cif.read_string(CCD_BLOCKS[ccd_code])[0]
     except KeyError:
         return
     ccd_reader_result = ccd_reader._parse_pdb_mmcif(ccd_block, sanitize=True)

{pxmeter-0.1.1 → pxmeter-0.1.3}/pxmeter/mapping.py

@@ -924,15 +924,28 @@ class MappingCIF:
         # This step will be change the res_id, res_name, atom_name in self.model_struct.atom_array
         self._align_model_lig_atom_to_ref(self.model_struct, model_to_ref_atom_mapping)
 
+        # Re-order model struct by res_id for each chain
+        order = []
+        for chain_id in np.unique(self.model_struct.uni_chain_id):
+            chain_mask = np.where(self.model_struct.uni_chain_id == chain_id)[0]
+            # Remove unmapped ligands
+            valid_chain_mask = chain_mask[
+                ~(
+                    (
+                        (self.model_struct.atom_array.res_name[chain_mask] == ".")
+                        & (self.model_struct.atom_array.atom_name[chain_mask] == ".")
+                    )
+                    | (
+                        self.model_struct.atom_array.res_id[chain_mask] < 0
+                    )  # -1 for unmapped residues
+                )
+            ]
+            res_ids = self.model_struct.atom_array.res_id[valid_chain_mask]
+            order.extend(valid_chain_mask[np.argsort(res_ids)])
+
         # Remove unmapped ligand of model and reset unique atom id
         self.model_struct = self.model_struct.select_substructure(
-            ~(
-                (
-                    (self.model_struct.atom_array.res_name == ".")
-                    & (self.model_struct.atom_array.atom_name == ".")
-                )
-                | (self.model_struct.atom_array.res_id < 0)  # -1 for unmapped residues
-            ),
+            order,
             reset_uni_id=True,
         )
 

pxmeter-0.1.3/pxmeter/metrics/clashes.py

@@ -0,0 +1,88 @@
+# Copyright 2025 ByteDance and/or its affiliates.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#      http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+from typing import Sequence
+
+import numpy as np
+from biotite.structure import AtomArray
+from biotite.structure.info.radii import vdw_radius_single
+from scipy.spatial import KDTree
+
+
+def check_clashes_by_vdw(
+    atom_array: AtomArray,
+    query_mask: Sequence[bool] = None,
+    vdw_scale_factor: float = 0.5,
+) -> list[tuple[int, int]]:
+    """
+    Check clashes between atoms in the given atom array.
+
+    Args:
+        atom_array (AtomArray): The atom array to check for clashes.
+        query_mask (bool, optional): A boolean mask to select atoms to check for clashes.
+                   If None, all atoms are checked.
+        vdw_scale_factor (float, optional): The scale factor to apply to the Van der Waals radii.
+                         Defaults to 0.5.
+
+    Returns:
+        list[tuple[int, int]]: A list of tuples representing the indices of atoms that are in clash.
+    """
+    if query_mask is None:
+        # query all atoms
+        query_mask = np.ones(len(atom_array), dtype=bool)
+    elif not np.any(query_mask):
+        # no query atoms, return empty list
+        return []
+
+    if query_mask is None:
+        query_mask = np.ones(len(atom_array), dtype=bool)
+
+    query_idx_in_ref = np.where(query_mask)[0]
+
+    vdw_radii = np.array([vdw_radius_single(e) for e in atom_array.element])
+    query_vdw_radii = vdw_radii[query_mask]
+
+    clashes = []
+    query_tree = KDTree(atom_array.coord)
+    for query_idx, nbs_idx in enumerate(
+        query_tree.query_ball_point(atom_array.coord[query_mask], r=3.0)
+    ):
+        query_bonds, _query_bond_types = atom_array.bonds.get_bonds(
+            query_idx_in_ref[query_idx]
+        )
+        query_vdw = query_vdw_radii[query_idx]
+        if query_vdw is None:
+            # undefined vdw for elem, use 1.7 as "C"
+            query_vdw = vdw_radius_single("C")
+
+        for nb_idx in nbs_idx:
+            if query_idx_in_ref[query_idx] == nb_idx:
+                # clash with self
+                continue
+
+            if nb_idx in query_bonds:
+                # clash with bonded atoms
+                continue
+
+            nb_vdw = vdw_radii[nb_idx]
+            if nb_vdw is None:
+                # undefined vdw for elem, use 1.7 as "C"
+                nb_vdw = vdw_radius_single("C")
+
+            dist = np.linalg.norm(
+                atom_array.coord[query_mask][query_idx] - atom_array.coord[nb_idx]
+            )
+            if dist < vdw_scale_factor * (query_vdw + nb_vdw):
+                clashes.append((query_idx_in_ref[query_idx], nb_idx))
+    return clashes

{pxmeter-0.1.1 → pxmeter-0.1.3/pxmeter.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: pxmeter
-Version: 0.1.1
+Version: 0.1.3
 Summary: PXMeter is a comprehensive toolkit for evaluating the quality of         structures generated by biomolecular structure prediction models.
 Author: Bytedance Inc.
 Author-email: ai4s-bio@bytedance.com

{pxmeter-0.1.1 → pxmeter-0.1.3}/pxmeter.egg-info/SOURCES.txt

@@ -26,6 +26,7 @@ pxmeter/data/struct.py
 pxmeter/data/utils.py
 pxmeter/data/writer.py
 pxmeter/metrics/__init__.py
+pxmeter/metrics/clashes.py
 pxmeter/metrics/lddt_metrics.py
 pxmeter/metrics/rmsd.py
 pxmeter/metrics/rmsd_metrics.py

{pxmeter-0.1.1 → pxmeter-0.1.3}/setup.py

@@ -20,7 +20,7 @@ with open("requirements.txt") as f:
 setup(
     name="pxmeter",
     python_requires=">=3.11",
-    version="0.1.1",
+    version="0.1.3",
     description="PXMeter is a comprehensive toolkit for evaluating the quality of \
         structures generated by biomolecular structure prediction models.",
     author="Bytedance Inc.",