bug fixes and additions to correctly handle PDB (mm)cif files.

molssi-seamm · May 7, 2022 · 2233c4f · 2233c4f
1 parent 335c4eb
commit 2233c4f
Show file tree

Hide file tree

Showing 3 changed files with 48 additions and 31 deletions.
diff --git a/read_structure_step/formats/cif/cif.py b/read_structure_step/formats/cif/cif.py
@@ -149,7 +149,7 @@ def load_cif(
 
                     # Set the system name
                     if system_name is not None and system_name != "":
-                        lower_name = system_name.lower()
+                        lower_name = str(system_name).lower()
                         if "from file" in lower_name:
                             system.name = block_name
                         elif "file name" in lower_name:
@@ -159,11 +159,11 @@ def load_cif(
                         elif "empirical formula" in lower_name:
                             system.name = configuration.formula()[1]
                         else:
-                            system.name = system_name
+                            system.name = str(system_name)
 
                     # And the configuration name
                     if configuration_name is not None and configuration_name != "":
-                        lower_name = configuration_name.lower()
+                        lower_name = str(configuration_name).lower()
                         if "from file" in lower_name:
                             configuration.name = block_name
                         elif "file name" in lower_name:
@@ -173,7 +173,7 @@ def load_cif(
                         elif "empirical formula" in lower_name:
                             configuration.name = configuration.formula()[1]
                         else:
-                            configuration.name = configuration_name
+                            configuration.name = str(configuration_name)
                     logger.debug(f"   added system {system_db.n_systems}: {block_name}")
                 block_name = line[5:].strip()
                 lines = []
@@ -194,7 +194,7 @@ def load_cif(
 
             # Set the system name
             if system_name is not None and system_name != "":
-                lower_name = system_name.lower()
+                lower_name = str(system_name).lower()
                 if "from file" in lower_name:
                     system.name = block_name
                 elif "file name" in lower_name:
@@ -204,11 +204,11 @@ def load_cif(
                 elif "empirical formula" in lower_name:
                     system.name = configuration.formula()[1]
                 else:
-                    system.name = system_name
+                    system.name = str(system_name)
 
             # And the configuration name
             if configuration_name is not None and configuration_name != "":
-                lower_name = configuration_name.lower()
+                lower_name = str(configuration_name).lower()
                 if "from file" in lower_name:
                     configuration.name = block_name
                 elif "file name" in lower_name:
@@ -218,4 +218,4 @@ def load_cif(
                 elif "empirical formula" in lower_name:
                     configuration.name = configuration.formula()[1]
                 else:
-                    configuration.name = configuration_name
+                    configuration.name = str(configuration_name)
diff --git a/read_structure_step/formats/cif/mmcif.py b/read_structure_step/formats/cif/mmcif.py
@@ -138,19 +138,26 @@ def load_mmcif(
                     block_name = line[5:].strip()
                 else:
                     structure_no += 1
-                    if structure_no > 1:
-                        if subsequent_as_configurations:
-                            configuration = system.create_configuration()
-                        else:
-                            system = system_db.create_system()
-                            configuration = system.create_configuration()
+                    # Check for NMR ensemble
+                    text = "\n".join(lines)
+                    if "_pdbx_nmr_ensemble.conformers_submitted_total_number" in text:
+                        system = system_db.create_system()
+                        system.from_mmcif_text(text)
+                    else:
+                        if structure_no > 1:
+                            if subsequent_as_configurations:
+                                configuration = system.create_configuration()
+                            else:
+                                system = system_db.create_system()
+                                configuration = system.create_configuration()
+
+                            configuration.from_mmcif_text(text)
 
-                    configuration.from_mmcif_text("\n".join(lines))
                     logger.debug(f"   added system {system_db.n_systems}: {block_name}")
 
                     # Set the system name
                     if system_name is not None and system_name != "":
-                        lower_name = system_name.lower()
+                        lower_name = str(system_name).lower()
                         if "from file" in lower_name:
                             system.name = block_name
                         elif "file name" in lower_name:
@@ -160,11 +167,11 @@ def load_mmcif(
                         elif "empirical formula" in lower_name:
                             system.name = configuration.formula()[1]
                         else:
-                            system.name = system_name
+                            system.name = str(system_name)
 
                     # And the configuration name
                     if configuration_name is not None and configuration_name != "":
-                        lower_name = configuration_name.lower()
+                        lower_name = str(configuration_name).lower()
                         if "from file" in lower_name:
                             configuration.name = block_name
                         elif "file name" in lower_name:
@@ -174,7 +181,7 @@ def load_mmcif(
                         elif "empirical formula" in lower_name:
                             configuration.name = configuration.formula()[1]
                         else:
-                            configuration.name = configuration_name
+                            configuration.name = str(configuration_name)
                     logger.debug(f"   added system {system_db.n_systems}: {block_name}")
                 block_name = line[5:].strip()
                 lines = []
@@ -183,19 +190,26 @@ def load_mmcif(
         if len(lines) > 0:
             # The last block just ends at the end of the file
             structure_no += 1
-            if structure_no > 1:
-                if subsequent_as_configurations:
-                    configuration = system.create_configuration()
-                else:
-                    system = system_db.create_system()
-                    configuration = system.create_configuration()
+            # Check for NMR ensemble
+            text = "\n".join(lines)
+            if "_pdbx_nmr_ensemble.conformers_submitted_total_number" in text:
+                system = system_db.create_system()
+                system.from_mmcif_text(text)
+            else:
+                if structure_no > 1:
+                    if subsequent_as_configurations:
+                        configuration = system.create_configuration()
+                    else:
+                        system = system_db.create_system()
+                        configuration = system.create_configuration()
+
+                    configuration.from_mmcif_text(text)
 
-            configuration.from_mmcif_text("\n".join(lines))
             logger.debug(f"   added system {system_db.n_systems}: {block_name}")
 
             # Set the system name
             if system_name is not None and system_name != "":
-                lower_name = system_name.lower()
+                lower_name = str(system_name).lower()
                 if "from file" in lower_name:
                     system.name = block_name
                 elif "file name" in lower_name:
@@ -205,11 +219,11 @@ def load_mmcif(
                 elif "empirical formula" in lower_name:
                     system.name = configuration.formula()[1]
                 else:
-                    system.name = system_name
+                    system.name = str(system_name)
 
             # And the configuration name
             if configuration_name is not None and configuration_name != "":
-                lower_name = configuration_name.lower()
+                lower_name = str(configuration_name).lower()
                 if "from file" in lower_name:
                     configuration.name = block_name
                 elif "file name" in lower_name:
@@ -219,4 +233,4 @@ def load_mmcif(
                 elif "empirical formula" in lower_name:
                     configuration.name = configuration.formula()[1]
                 else:
-                    configuration.name = configuration_name
+                    configuration.name = str(configuration_name)
diff --git a/read_structure_step/read_structure.py b/read_structure_step/read_structure.py
@@ -181,6 +181,9 @@ def run(self):
             )
 
             # Finish the output
+            system, configuration = self.get_system_configuration(
+                P, structure_handling=False
+            )
             if configuration.periodicity == 3:
                 space_group = configuration.symmetry.group
                 if space_group == "":
@@ -200,7 +203,7 @@ def run(self):
                 printer.important(
                     __(
                         "\n    Created a molecular structure with "
-                        "{configuration.n_atoms} atoms."
+                        f"{configuration.n_atoms} atoms."
                         f"\n           System name = {system.name}"
                         f"\n    Configuration name = {configuration.name}",
                         indent=4 * " ",