Chimie Théorique » scripts_chimie4psmn » DockOnSurf

"""Module for the conversion between atomic coordinates files and objects

functions:

confs_to_mol_list: Converts the conformers inside a rdkit mol object to a list

    of separate mol objects.

rdkit_mol_to_ase_atoms: Converts a rdkit mol object into ase Atoms object.

adapt_format: Converts the coordinate files into a required library object type.

read_coords: Reads the atomic coordinates resulting from finished calculations.

"""

import logging

import rdkit.Chem.AllChem as Chem

logger = logging.getLogger('DockOnSurf')

def confs_to_mol_list(mol: Chem.rdchem.Mol, idx_lst=None):

    """Converts the conformers inside a rdkit mol object to a list of

    separate mol objects.

    @param mol: rdkit mol object containing at least one conformer.

    @param idx_lst: list of conformer indices to be considered. If not passed,

        all conformers are considered.

    @return: list of separate mol objects.

    """

    if idx_lst is None:

        idx_lst = list(range(mol.GetNumConformers()))

    return [Chem.MolFromMolBlock(Chem.MolToMolBlock(mol, confId=int(idx)),

                                 removeHs=False) for idx in idx_lst]

def rdkit_mol_to_ase_atoms(mol: Chem.rdchem.Mol):

    """Converts a rdkit mol object into ase Atoms object.

    @param mol: rdkit mol object containing only one conformer.

    @return ase.Atoms: ase Atoms object with the same coordinates.

    """

    from ase import Atoms

    if mol.GetNumConformers() > 1:

        logger.warning('A mol object with multiple conformers is parsed, '

                       'converting to Atoms only the first conformer')

    symbols = [atm.GetSymbol() for atm in mol.GetAtoms()]

    positions = mol.GetConformer(0).GetPositions()

    return Atoms(symbols=symbols, positions=positions)

def adapt_format(requirement, coord_file):

    """Converts the coordinate files into a required library object type.

    Depending on the library required to use and the file type, it converts the

    coordinate file into a library-workable object.

    @param requirement: str, the library for which the conversion should be

    made. Accepted values: 'ase', 'rdkit'.

    @param coord_file: str, path to the coordinates file aiming to convert.

    Accepted file tyoes: 'xyz', 'mol'.

    @return: an object the required library can work with.

    """

    import ase.io

    from ase.io.formats import filetype

    req_vals = ['rdkit', 'ase']

    file_type_vals = ['xyz', 'mol']

    lib_err = f"The conversion to the '{requirement}' library object type" \

              f" has not yet been implemented"

    conv_info = f"Converted {coord_file} to {requirement} object type"

    fil_type_err = f'The {filetype(coord_file)} file formnat is not supported'

    if requirement not in req_vals:

        logger.error(lib_err)

        raise NotImplementedError(lib_err)

    if filetype(coord_file) not in file_type_vals:

        logger.error(fil_type_err)

        raise NotImplementedError(fil_type_err)

    if requirement == 'rdkit':

        if filetype(coord_file) == 'xyz':

            from xyz2mol import xyz2mol

            ase_atms = ase.io.read(coord_file)

            atomic_nums = ase_atms.get_atomic_numbers().tolist()

            xyz_coordinates = ase_atms.positions.tolist()

            # TODO Add routine to read charge

            rd_mol_obj = xyz2mol(atomic_nums, xyz_coordinates, charge=0)

            logger.debug(conv_info)

            return rd_mol_obj

        elif filetype(coord_file) == 'mol':

            from rdkit.Chem import MolFromMolFile

            logger.debug(conv_info)

            return MolFromMolFile(coord_file, removeHs=False)

    if requirement == 'ase':

        logger.debug(conv_info)

        return ase.io.read(coord_file)

def read_coords(code, run_type, req):

    """Reads the atomic coordinates resulting from finished calculations.

    Given a run_type ('isolated', 'screening' or 'refinement') directory

    containing different subdirectories with finished calculations in every

    subdirectory, it reads the coordinates inside each subdirectory according to

    the specified code and returns a list of objects required.

    @param code: the code that produced the coordinates out of a calculation.

    @param run_type: the type of calculation (and also the name of the folder)

                     containing the calculation subdirectories.

    @param req: The required object type to make the list of.

    @return: list of collection-of-atoms objects. (rdkit.Mol, ase.Atoms, etc.)

    """

    import os

    if code == 'cp2k':

        pattern = '-pos-1.xyz'

    else:

        pattern = ''

    return [adapt_format(req, f'{run_type}/{conf}/{fil}')

            for conf in os.listdir(run_type)

            for fil in os.listdir(f"{run_type}/{conf}") if pattern in fil]