gemmi-protools 0.1.2__py3-none-any.whl → 0.1.3__py3-none-any.whl

gemmi_protools/reader.py

@@ -2,7 +2,6 @@
 @Author: Luo Jiejian
 """
 import pathlib
-import re
 import string
 import warnings
 from copy import deepcopy
@@ -10,6 +9,7 @@ from typing import Union, Optional, List
 
 import gemmi
 import numpy as np
+from Bio.PDB.Polypeptide import nucleic_letters_3to1_extended, protein_letters_3to1_extended
 from typeguard import typechecked
 
 from .cif_opts import _cif_block_for_output, _is_cif
@@ -19,6 +19,10 @@ from .parser import (_assign_digital_entity_names, _ent_from_structure,
 from .pdb_opts import _compound_source_string, _is_pdb
 from .struct_info import Info
 
+nucleic_letters_3to1_extended["DN"] = "N"
+nucleic_letters_3to1_extended["N"] = "N"
+protein_letters_3to1_extended["UNK"] = "X"
+
 
 class StructureParser(object):
     """
@@ -42,6 +46,27 @@ class StructureParser(object):
         self.INFO.from_gemmi_structure_infomap(self.STRUCT.info)
         self.ENTITY = _ent_from_structure(self.STRUCT)
         self.update_entity()
+        self.update_full_sequences()
+
+    def update_full_sequences(self):
+        for ent_idx, ent in enumerate(self.STRUCT.entities):
+            # get full sequence
+            full_seq = ent.full_sequence
+
+            # when missing, construct from Residues
+            if not full_seq:
+                sel_ch_id = None
+                sel_ch_len = 0
+                for ch_id, ent_id in self.ENTITY.polymer2eid.items():
+                    if ent_id == ent.name:
+                        cur_len = len(self.polymer_sequences[ch_id])
+                        if cur_len > sel_ch_len:
+                            sel_ch_id = ch_id
+                            sel_ch_len = cur_len
+
+                if sel_ch_id is not None and sel_ch_len > 0:
+                    full_seq = [r.name for r in self.STRUCT[0][sel_ch_id].get_polymer() if not r.is_water()]
+                    self.STRUCT.entities[ent_idx].full_sequence = full_seq
 
     @typechecked
     def load_from_file(self, path: Union[str, pathlib.PosixPath]):
@@ -56,6 +81,7 @@ class StructureParser(object):
         self.STRUCT, self.ENTITY = struct, entity
         self.INFO.from_gemmi_structure_infomap(self.STRUCT.info)
         self.update_entity()
+        self.update_full_sequences()
 
     @typechecked
     def to_pdb(self, outfile: str, write_minimal_pdb=False):
@@ -131,7 +157,7 @@ class StructureParser(object):
             for chain in model:
                 ct = cts.get(chain.name, "other")
                 if ct != "other":
-                    out[chain.name] = re.sub("-", "", chain.get_polymer().make_one_letter_sequence())
+                    out[chain.name] = self.make_one_letter_sequence(chain.name)
         return out
 
     @property
@@ -152,12 +178,28 @@ class StructureParser(object):
                                                 for r in chain.get_polymer()], dtype=id_type)
         return out
 
-    @property
-    def chain_residues(self):
+    def chain_residues(self, polymer_only=True, with_water=False):
+        """
+        :param polymer_only, bool
+        :param with_water:
+        :return: dict of Three-letter codes of chain residues
+        """
+
         out = dict()
         for model in self.STRUCT:
             for chain in model:
-                out[chain.name] = [r.name for r in chain]
+                res_codes = []
+                for r in chain:
+                    if r.is_water():
+                        if with_water:
+                            res_codes.append(r.name)
+                    else:
+                        if polymer_only:
+                            if r.entity_type.name == "Polymer":
+                                res_codes.append(r.name)
+                        else:
+                            res_codes.append(r.name)
+                out[chain.name] = res_codes
         return out
 
     def update_entity(self):
@@ -369,3 +411,15 @@ class StructureParser(object):
             return np.array(coord, dtype=np.float32), np.array(atom_id, dtype=id_type)
         else:
             return np.empty(shape=(0, 3), dtype=np.float32), np.array(atom_id, dtype=id_type)
+
+    def make_one_letter_sequence(self, chain_id):
+        c_type = self.chain_types[chain_id]
+        residues = self.chain_residues(polymer_only=True, with_water=False)[chain_id]
+
+        if c_type == "protein":
+            one_letter_code = "".join([protein_letters_3to1_extended.get(r, "X") for r in residues])
+        elif c_type in ["dna", "rna"]:
+            one_letter_code = "".join([nucleic_letters_3to1_extended.get(r, "N") for r in residues])
+        else:
+            one_letter_code = ""
+        return one_letter_code

{gemmi_protools-0.1.2.dist-info → gemmi_protools-0.1.3.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: gemmi_protools
-Version: 0.1.2
+Version: 0.1.3
 Summary: An Enhanced tool to process PDB structures based on Gemmi
 Author-email: Luo Jiejian <luojiejian12@mails.ucas.ac.cn>
 License-Expression: MIT

{gemmi_protools-0.1.2.dist-info → gemmi_protools-0.1.3.dist-info}/RECORD

@@ -7,10 +7,10 @@ gemmi_protools/parse_pdb_header.py,sha256=UOGMsE3-d3APhO7zaAEE0NT31n-iqt55VpDh_R
 gemmi_protools/parser.py,sha256=QIJCOfK8FaFbLMvBG82zTOAjIvQJcf2WRwuFSxj4zvc,8982
 gemmi_protools/pdb_opts.py,sha256=NbXLDNNVF7tuG_bUM0Infylf5aYnOCP2Pd-ndqm5bK4,5652
 gemmi_protools/ppi.py,sha256=nRzRWv28SDjVt6hMShRL_QYKFsBO1xA5jSGIQrN0JBg,2313
-gemmi_protools/reader.py,sha256=0VjMxOogqB1dccucQU2I703W5Ro4JYBakE2eXftUTPU,13194
+gemmi_protools/reader.py,sha256=4xuK7UmJVc6Nt9YhYS1FPsi5PL4QwZHilk8s3V-jf0Y,15499
 gemmi_protools/struct_info.py,sha256=9nBj1Zer03S8_Wks7L7uRlc9PlbfCKzoaT32pKR58X8,2769
-gemmi_protools-0.1.2.dist-info/licenses/LICENSE,sha256=JuQvKcgj6n11y5y6nXr9rABv3gJSswc4eTCd5WZBtSY,1062
-gemmi_protools-0.1.2.dist-info/METADATA,sha256=XTB_XuZpn611rYsLD2GMq6qd_bwm-TDBtGcNr-NROzE,567
-gemmi_protools-0.1.2.dist-info/WHEEL,sha256=CmyFI0kx5cdEMTLiONQRbGQwjIoR1aIYB7eCAQ4KPJ0,91
-gemmi_protools-0.1.2.dist-info/top_level.txt,sha256=P12mYJi5O5EKIn5u-RFaWxuix431CgLacSRD7rBid_U,15
-gemmi_protools-0.1.2.dist-info/RECORD,,
+gemmi_protools-0.1.3.dist-info/licenses/LICENSE,sha256=JuQvKcgj6n11y5y6nXr9rABv3gJSswc4eTCd5WZBtSY,1062
+gemmi_protools-0.1.3.dist-info/METADATA,sha256=ymS32ZVWiYY7LDzrQkEqiatqY2qZBqenf3X0BvX5DT4,567
+gemmi_protools-0.1.3.dist-info/WHEEL,sha256=CmyFI0kx5cdEMTLiONQRbGQwjIoR1aIYB7eCAQ4KPJ0,91
+gemmi_protools-0.1.3.dist-info/top_level.txt,sha256=P12mYJi5O5EKIn5u-RFaWxuix431CgLacSRD7rBid_U,15
+gemmi_protools-0.1.3.dist-info/RECORD,,