Mol2, SDF and XYZ File Parsers

.gitignore

@@ -55,3 +55,4 @@ biotite.egg-info
 
 # Ignore fuse_hidden files on Linux systems
 *.fuse_hidden*
+

src/biotite/structure/io/ctab.py

@@ -11,6 +11,8 @@
 __author__ = "Patrick Kunzmann"
 __all__ = ["read_structure_from_ctab", "write_structure_to_ctab"]
 
+import traceback
+
 import warnings
 import numpy as np
 from biotite.structure.error import BadStructureError
@@ -61,6 +63,8 @@ def read_structure_from_ctab(ctab_lines):
     .. footbibliography::
     """
     n_atoms, n_bonds = _get_counts(ctab_lines[0])
+
+
     atom_lines = ctab_lines[1 : 1 + n_atoms]
     bond_lines = ctab_lines[1 + n_atoms : 1 + n_atoms + n_bonds]
 
@@ -163,5 +167,6 @@ def write_structure_to_ctab(atoms, default_bond_type=BondType.ANY):
 
 
 def _get_counts(counts_line):
+
     elements = counts_line.split()
     return int(elements[0]), int(elements[1])

src/biotite/structure/io/general.py

@@ -78,7 +78,8 @@ def load_structure(file_path, template=None, **kwargs):
             # Stack containing only one model -> return as atom array
             return array[0]
         else:
-            return array
+            return array                    
+
     elif suffix == ".cif" or suffix == ".pdbx":
         from .pdbx import PDBxFile, get_structure
         file = PDBxFile.read(file_path)
@@ -115,12 +116,33 @@ def load_structure(file_path, template=None, **kwargs):
             return array[0]
         else:
             return array
-    elif suffix == ".mol" or suffix == ".sdf":
+    elif suffix == ".mol":
         from .mol import MOLFile
         file = MOLFile.read(file_path)
         array = file.get_structure(**kwargs)
         # MOL files only contain a single model
         return array
+    elif suffix == ".sdf":
+        from .sdf import SDFile
+        file = SDFile.read(file_path)
+        array = file.get_structure(**kwargs)
+        #  SDFile automaticall detects if to return        
+        #  AomArray or AtomArrayStack
+        return array          
+    elif suffix == ".xyz":
+        from .xyz import XYZFile
+        file = XYZFile.read(file_path)
+        array = file.get_structure(**kwargs)
+        # XYZFile automaticall detects if to return
+        #  AomArray or AtomArrayStack
+        return array
+    elif suffix == ".mol2":
+        from .mol2 import MOL2File
+        file = MOL2File.read(file_path)
+        array = file.get_structure(**kwargs)
+        # MOL2File automaticall detects if to return        
+        #  AomArray or AtomArrayStack
+        return array        
     elif suffix in [".trr", ".xtc", ".tng", ".dcd", ".netcdf"]:
         if template is None:
             raise TypeError("Template must be specified for trajectory files")
@@ -204,11 +226,26 @@ def save_structure(file_path, array, **kwargs):
         file = NpzFile()
         file.set_structure(array, **kwargs)
         file.write(file_path)
-    elif suffix == ".mol" or suffix == ".sdf":
+    elif suffix == ".mol":
         from .mol import MOLFile
         file = MOLFile()
         file.set_structure(array, **kwargs)
         file.write(file_path)
+    elif suffix == ".sdf":
+        from .sdf import SDFile
+        file = SDFile()
+        file.set_structure(array, **kwargs)
+        file.write(file_path)  
+    elif suffix == ".xyz":
+        from .xyz import XYZFile
+        file = XYZFile()
+        file.set_structure(array, **kwargs)
+        file.write(file_path) 
+    elif suffix == ".mol2":
+        from .mol2 import MOL2File
+        file = MOL2File()
+        file.set_structure(array, **kwargs)
+        file.write(file_path)                      
     elif suffix in [".trr", ".xtc", ".tng", ".dcd", ".netcdf"]:
         from .trr import TRRFile
         from .xtc import XTCFile

src/biotite/structure/io/mol/file.py

@@ -91,7 +91,8 @@ def get_header(self):
         program : str
             The program name.
         time : datetime
-            The time of file creation.
+            The time of file creation. Returns None in this field if not
+            able to parse according entry in MOLFile as datetime string.
         dimensions : str
             Dimensional codes.
         scaling_factors : str
@@ -106,8 +107,21 @@ def get_header(self):
         mol_name        = self.lines[0].strip()
         initials        = self.lines[1][0:2].strip()
         program         = self.lines[1][2:10].strip()
-        time            = datetime.datetime.strptime(self.lines[1][10:20],
-                                                     DATE_FORMAT)
+
+
+        # sometimes the string can not be interpreted as datetime
+        # in those cases instead of failing simply warn the user
+        time = None       
+        try:
+            time        = datetime.datetime.strptime(
+                self.lines[1][10:20],
+                DATE_FORMAT
+            )
+        except:
+            line = format(self.lines[1][10:20])
+            warn("{} could not be interpreted as datetime".format(line))
+
+
         dimensions      = self.lines[1][20:22].strip()
         scaling_factors = self.lines[1][22:34].strip()
         energy          = self.lines[1][34:46].strip()
@@ -132,7 +146,7 @@ def set_header(self, mol_name, initials="", program="", time=None,
         program : str, optional
             The program name. Maximum length is 8.
         time : datetime or date, optional
-            The time of file creation.
+            The time of file creation, if none uses current time.
         dimensions : str, optional
             Dimensional codes. Maximum length is 2.
         scaling_factors : str, optional

src/biotite/structure/io/mol2/__init__.py

@@ -0,0 +1,16 @@
+# This source code is part of the Biotite package and is distributed
+# under the 3-Clause BSD License. Please see 'LICENSE.rst' for further
+# information.
+
+"""
+The MOL format is used to depict atom positions and bonds for small
+molecules.
+This subpackage is used for reading and writing an :class:`AtomArray` or
+an :class:`AtomArrayStack` for a file containing multiple models in this format.
+"""
+
+__name__ = "biotite.structure.io.mol2"
+__author__ = "Benjamin E. Mayer"
+
+from .file import *
+from .convert import *

src/biotite/structure/io/mol2/convert.py

@@ -0,0 +1,99 @@
+# This source code is part of the Biotite package and is distributed
+# under the 3-Clause BSD License. Please see 'LICENSE.rst' for further
+# information.
+
+__name__ = "biotite.structure.io.mol2"
+__author__ = "Benjamin E. Mayer"
+__all__ = [
+    "get_structure", "set_structure",
+    "get_charges", "set_charges"
+]
+
+
+
+def get_structure(mol2_file):
+    """
+    Get an :class:`AtomArray` from the MOL2 File.
+
+    Ths function is a thin wrapper around
+    :meth:`MOL2File.get_structure()`.
+
+    Parameters
+    ----------
+    mol2_file : MOL2File
+        The MOL2File.
+
+    Returns
+    -------
+    array : AtomArray, AtomArrayStack
+        Return an AtomArray or AtomArrayStack containing the structure or
+        structures depending on if file contains single or multiple models.        
+        If something other then `NO_CHARGE` is set in the charge_type field
+        of the according mol2 file, the AtomArray or AtomArrayStack will 
+        contain the charge field.
+    """
+    return mol2_file.get_structure()
+
+
+def set_structure(mol2_file, atoms):
+    """
+    Set the :class:`AtomArray` for the MOL2 File.
+
+    Ths function is a thin wrapper around
+    :meth:`MOL2File.set_structure(atoms)`.
+
+    Parameters
+    ----------
+    mol2_file : MOL2File
+        The XYZ File.
+    array : AtomArray
+        The array to be saved into this file.
+        Must have an associated :class:`BondList`.
+        If charge field set this is used for storage within the according
+        MOL2 charge column.
+    """
+    mol2_file.set_structure(atoms)
+
+
+
+def get_charges(mol2_file):
+    """
+    Get an ndarray containing the partial charges from the MOL2File
+
+    Ths function is a thin wrapper around
+    :meth:`MOL2File.get_charges()`.
+
+    Parameters
+    ----------
+    xyz_file : XYZFile
+        The XYZ File.
+
+    Returns
+    -------
+    array : AtomArray
+        This :class:`AtomArray` contains the optional ``charge``
+        annotation and has an associated :class:`BondList`.
+        All other annotation categories, except ``element`` are
+        empty.
+    """
+    return mol2_file.get_charges()    
+
+
+def set_charges(mol2_file, charges):
+    """
+    Set the partial charges in the MOL2File to an ndarray
+    specified as parameter here.
+
+    Ths function is a thin wrapper around
+    :meth:`MOL2File.set_charges(charges)`.
+
+    Parameters
+    ----------
+    mol2_file: MOL2File
+        The MOL2File
+    charges: ndarray
+        A ndarray containing data with `float` type to be written as 
+        partial charges.           
+
+    """
+    return mol2_file.set_charges(charges)