dimorphite_dl.py

# Copyright 2018 Jacob D. Durrant
#
# 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.

"""
This script identifies and enumerates the possible protonation sites of SMILES
strings.
"""

from __future__ import print_function
import copy
import os
import argparse
import sys

try:
    # Python2
    from StringIO import StringIO
except ImportError:
    # Python3
    from io import StringIO

# Always let the user know a help file is available.
print("\nFor help, use: python dimorphite_dl.py --help")

# And always report citation information.
print("\nIf you use Dimorphite-DL in your research, please cite:")
print("Ropp PJ, Kaminsky JC, Yablonski S, Durrant JD (2019) Dimorphite-DL: An")
print("open-source program for enumerating the ionization states of drug-like small")
print("molecules. J Cheminform 11:14. doi:10.1186/s13321-019-0336-9.\n")

try:
    import rdkit
    from rdkit import Chem
    from rdkit.Chem import AllChem
except:
    msg = "Dimorphite-DL requires RDKit. See https://www.rdkit.org/"
    print(msg)
    raise Exception(msg)

def main(params=None):
    """The main definition run when you call the script from the commandline.

    :param params: The parameters to use. Entirely optional. If absent,
                   defaults to None, in which case argments will be taken from
                   those given at the command line.
    :param params: dict, optional
    :return: Returns a list of the SMILES strings return_as_list parameter is
             True. Otherwise, returns None.
    """

    parser = ArgParseFuncs.get_args()
    args = vars(parser.parse_args())

    # Add in any parameters in params.
    if params is not None:
        for k, v in params.items():
            args[k] = v

    # If being run from the command line, print out all parameters.
    if __name__ == "__main__":
        print("\nPARAMETERS:\n")
        for k in sorted(args.keys()):
            print(k.rjust(13) + ": " + str(args[k]))
        print("")

    if args["test"]:
        # Run tests.
        TestFuncs.test()
    else:
        # Run protonation
        if "output_file" in args and args["output_file"] is not None:
            # An output file was specified, so write to that.
            with open(args["output_file"], "w") as file:
                for protonated_smi in Protonate(args):
                    file.write(protonated_smi + "\n")
        elif "return_as_list" in args and args["return_as_list"] == True:
            return list(Protonate(args))
        else:
            # No output file specified. Just print it to the screen.
            for protonated_smi in Protonate(args):
                print(protonated_smi)

class MyParser(argparse.ArgumentParser):
    """Overwrite default parse so it displays help file on error. See
    https://stackoverflow.com/questions/4042452/display-help-message-with-python-argparse-when-script-is-called-without-any-argu"""

    def error(self, message):
        """Overwrites the default error message.

        :param message: The default error message.
        """

        self.print_help()
        msg = "ERROR: %s\n\n" % message
        print(msg)
        raise Exception(msg)

    def print_help(self, file=None):
        """Overwrite the default print_help function

        :param file: Output file, defaults to None
        """

        print("")

        if file is None:
            file = sys.stdout
        self._print_message(self.format_help(), file)
        print("""
examples:
  python dimorphite_dl.py --smiles_file sample_molecules.smi
  python dimorphite_dl.py --smiles "CCC(=O)O" --min_ph -3.0 --max_ph -2.0
  python dimorphite_dl.py --smiles "CCCN" --min_ph -3.0 --max_ph -2.0 --output_file output.smi
  python dimorphite_dl.py --smiles_file sample_molecules.smi --pka_precision 2.0 --label_states
  python dimorphite_dl.py --test""")
        print("")

class ArgParseFuncs:
    """A namespace for storing functions that are useful for processing
    command-line arguments. To keep things organized."""

    @staticmethod
    def get_args():
        """Gets the arguments from the command line.

        :return: A parser object.
        """

        parser = MyParser(description="Dimorphite 1.2: Creates models of " +
                                    "appropriately protonated small moleucles. " +
                                    "Apache 2.0 License. Copyright 2018 Jacob D. " +
                                    "Durrant.")
        parser.add_argument('--min_ph', metavar='MIN', type=float, default=6.4,
                            help='minimum pH to consider (default: 6.4)')
        parser.add_argument('--max_ph', metavar='MAX', type=float, default=8.4,
                            help='maximum pH to consider (default: 8.4)')
        parser.add_argument('--pka_precision', metavar='PRE', type=float, default=1.0,
                            help='pKa precision factor (number of standard devations, default: 1.0)')
        parser.add_argument('--smiles', metavar='SMI', type=str,
                            help='SMILES string to protonate')
        parser.add_argument('--smiles_file', metavar="FILE", type=str,
                            help='file that contains SMILES strings to protonate')
        parser.add_argument('--output_file', metavar="FILE", type=str,
                            help='output file to write protonated SMILES (optional)')
        parser.add_argument('--label_states', action="store_true",
                            help='label protonated SMILES with target state ' + \
                                '(i.e., "DEPROTONATED", "PROTONATED", or "BOTH").')
        parser.add_argument('--test', action="store_true",
                            help='run unit tests (for debugging)')

        return parser

    @staticmethod
    def clean_args(args):
        """Cleans and normalizes input parameters

        :param args: A dictionary containing the arguments.
        :type args: dict
        :raises Exception: No SMILES in params.
        """

        defaults = {'min_ph' : 6.4,
                    'max_ph' : 8.4,
                    'pka_precision' : 1.0,
                    'label_states' : False,
                    'test' : False}

        for key in defaults:
            if key not in args:
                args[key] = defaults[key]

        keys = list(args.keys())
        for key in keys:
            if args[key] is None:
                del args[key]

        if not "smiles" in args and not "smiles_file" in args:
            msg = "Error: No SMILES in params. Use the -h parameter for help."
            print(msg)
            raise Exception(msg)

        # If the user provides a smiles string, turn it into a file-like StringIO
        # object.
        if "smiles" in args:
            if isinstance(args["smiles"], str):
                args["smiles_file"]  = StringIO(args["smiles"])

        args["smiles_and_data"] = LoadSMIFile(args["smiles_file"])

        return args

class UtilFuncs:
    """A namespace to store functions for manipulating mol objects. To keep
    things organized."""

    @staticmethod
    def neutralize_mol(mol):
        """All molecules should be neuralized to the extent possible. The user
        should not be allowed to specify the valence of the atoms in most cases.

        :param rdkit.Chem.rdchem.Mol mol: The rdkit Mol objet to be neutralized.
        :return: The neutralized Mol object.
        """

        # Get the reaction data
        rxn_data = [
            ['[Ov1-1:1]', '[Ov2+0:1]-[H]'],  # To handle O- bonded to only one atom (add hydrogen).
            ['[#7v4+1:1]-[H]', '[#7v3+0:1]'],  # To handle N+ bonded to a hydrogen (remove hydrogen).
            ['[Ov2-:1]', '[Ov2+0:1]'],  # To handle O- bonded to two atoms. Should not be Negative.
            ['[#7v3+1:1]', '[#7v3+0:1]'],  # To handle N+ bonded to three atoms. Should not be positive.
            ['[#7v2-1:1]', '[#7+0:1]-[H]'],  # To handle N- Bonded to two atoms. Add hydrogen.
            # ['[N:1]=[N+0:2]=[N:3]-[H]', '[N:1]=[N+1:2]=[N+0:3]-[H]'],  # To
            # handle bad azide. Must be protonated. (Now handled elsewhere, before
            # SMILES converted to Mol object.)
            ['[H]-[N:1]-[N:2]#[N:3]', '[N:1]=[N+1:2]=[N:3]-[H]']  # To handle bad azide. R-N-N#N should be R-N=[N+]=N
        ]

        # Add substructures and reactions (initially none)
        for i, rxn_datum in enumerate(rxn_data):
            rxn_data[i].append(Chem.MolFromSmarts(rxn_datum[0]))
            rxn_data[i].append(None)

        # Add hydrogens (respects valence, so incomplete).
        # Chem.calcImplicitValence(mol)
        mol.UpdatePropertyCache(strict=False)
        mol = Chem.AddHs(mol)

        while True:  # Keep going until all these issues have been resolved.
            current_rxn = None  # The reaction to perform.
            current_rxn_str = None

            for i, rxn_datum in enumerate(rxn_data):
                reactant_smarts, product_smarts, substruct_match_mol, rxn_placeholder = rxn_datum
                if mol.HasSubstructMatch(substruct_match_mol):
                    if rxn_placeholder is None:
                        current_rxn_str = reactant_smarts + '>>' + product_smarts
                        current_rxn = AllChem.ReactionFromSmarts(current_rxn_str)
                        rxn_data[i][3] = current_rxn  # Update the placeholder.
                    else:
                        current_rxn = rxn_data[i][3]
                    break

            # Perform the reaction if necessary
            if current_rxn is None:  # No reaction left, so break out of while loop.
                break
            else:
                mol = current_rxn.RunReactants((mol,))[0][0]
                mol.UpdatePropertyCache(strict=False)  # Update valences

        # The mols have been altered from the reactions described above, we need
        # to resanitize them. Make sure aromatic rings are shown as such This
        # catches all RDKit Errors. without the catchError and sanitizeOps the
        # Chem.SanitizeMol can crash the program.
        sanitize_string =  Chem.SanitizeMol(
            mol,
            sanitizeOps=rdkit.Chem.rdmolops.SanitizeFlags.SANITIZE_ALL,
            catchErrors = True
        )

        return mol if sanitize_string.name == "SANITIZE_NONE" else None

    @staticmethod
    def convert_smiles_str_to_mol(smiles_str):
        """Given a SMILES string, check that it is actually a string and not a
        None. Then try to convert it to an RDKit Mol Object.

        :param string smiles_str: The SMILES string.
        :return: A rdkit.Chem.rdchem.Mol object, or None if it is the wrong type or
            if it fails to convert to a Mol Obj
        """

        # Check that there are no type errors, ie Nones or non-string
        # A non-string type will cause RDKit to hard crash
        if smiles_str is None or type(smiles_str) is not str:
            return None

        # Try to fix azides here. They are just tricky to deal with.
        smiles_str = smiles_str.replace("N=N=N", "N=[N+]=N")
        smiles_str = smiles_str.replace("NN#N", "N=[N+]=N")

        # Now convert to a mol object. Note the trick that is necessary to
        # capture RDKit error/warning messages. See
        # https://stackoverflow.com/questions/24277488/in-python-how-to-capture-the-stdout-from-a-c-shared-library-to-a-variable
        stderr_fileno = sys.stderr.fileno()
        stderr_save = os.dup(stderr_fileno)
        stderr_pipe = os.pipe()
        os.dup2(stderr_pipe[1], stderr_fileno)
        os.close(stderr_pipe[1])

        mol = Chem.MolFromSmiles(smiles_str)

        os.close(stderr_fileno)
        os.close(stderr_pipe[0])
        os.dup2(stderr_save, stderr_fileno)
        os.close(stderr_save)

        # Check that there are None type errors Chem.MolFromSmiles has sanitize on
        # which means if there is even a small error in the SMILES (kekulize,
        # nitrogen charge...) then mol=None. ie.
        # Chem.MolFromSmiles("C[N]=[N]=[N]") = None this is an example of an
        # nitrogen charge error. It is cased in a try statement to be overly
        # cautious.

        return None if mol is None else mol

    @staticmethod
    def eprint(*args, **kwargs):
        """Error messages should be printed to STDERR. See
        https://stackoverflow.com/questions/5574702/how-to-print-to-stderr-in-python"""

        print(*args, file=sys.stderr, **kwargs)

class LoadSMIFile(object):
    """A generator class for loading in the SMILES strings from a file, one at
    a time."""

    def __init__(self, filename):
        """Initializes this class.

        :param filename: The filename or file object (i.e., StringIO).
        :type filename: str or StringIO
        """

        if type(filename) is str:
            # It's a filename
            self.f = open(filename, "r")
        else:
            # It's a file object (i.e., StringIO)
            self.f = filename

    def __iter__(self):
        """Returns this generator object.

        :return: This generator object.
        :rtype: LoadSMIFile
        """

        return self

    def __next__(self):
        """Ensure Python3 compatibility.

        :return: A dict, where the "smiles" key contains the canonical SMILES
                 string and the "data" key contains the remaining information
                 (e.g., the molecule name).
        :rtype: dict
        """

        return self.next()

    def next(self):
        """Get the data associated with the next line.

        :raises StopIteration: If there are no more lines left iin the file.
        :return: A dict, where the "smiles" key contains the canonical SMILES
                 string and the "data" key contains the remaining information
                 (e.g., the molecule name).
        :rtype: dict
        """

        line = self.f.readline()

        if line == "":
            # EOF
            self.f.close()
            raise StopIteration()
            return

        # Divide line into smi and data
        splits = line.split()
        if len(splits) != 0:
            # Generate mol object
            smiles_str = splits[0]

            # Convert from SMILES string to RDKIT Mol. This series of tests is
            # to make sure the SMILES string is properly formed and to get it
            # into a canonical form. Filter if failed.
            mol = UtilFuncs.convert_smiles_str_to_mol(smiles_str)
            if mol is None:
                UtilFuncs.eprint("WARNING: Skipping poorly formed SMILES string: " + line)
                return self.next()

            # Handle nuetralizing the molecules. Filter if failed.
            mol = UtilFuncs.neutralize_mol(mol)
            if mol is None:
                UtilFuncs.eprint("WARNING: Skipping poorly formed SMILES string: " + line)
                return self.next()

            # Remove the hydrogens.
            try:
                mol = Chem.RemoveHs(mol)
            except:
                UtilFuncs.eprint("WARNING: Skipping poorly formed SMILES string: " + line)
                return self.next()

            if mol is None:
                UtilFuncs.eprint("WARNING: Skipping poorly formed SMILES string: " + line)
                return self.next()

            # Regenerate the smiles string (to standardize).
            new_mol_string = Chem.MolToSmiles(mol, isomericSmiles=True)

            return {
                "smiles": new_mol_string,
                "data": splits[1:]
            }
        else:
            # Blank line? Go to next one.
            return self.next()

class Protonate(object):
    """A generator class for protonating SMILES strings, one at a time."""

    def __init__(self, args):
        """Initialize the generator.

        :param args: A dictionary containing the arguments.
        :type args: dict
        """

        # Make the args an object variable variable.
        self.args = args

        # A list to store the protonated SMILES strings associated with a
        # single input model.
        self.cur_prot_SMI = []

        # Clean and normalize the args
        self.args = ArgParseFuncs.clean_args(args)

        # Load the substructures that can be protonated.
        self.subs = ProtSubstructFuncs.load_protonation_substructs_calc_state_for_ph(
            self.args["min_ph"], self.args["max_ph"], self.args["pka_precision"]
        )

    def __iter__(self):
        """Returns this generator object.

        :return: This generator object.
        :rtype: Protonate
        """

        return self

    def __next__(self):
        """Ensure Python3 compatibility.

        :return: A dict, where the "smiles" key contains the canonical SMILES
                 string and the "data" key contains the remaining information
                 (e.g., the molecule name).
        :rtype: dict
        """

        return self.next()

    def next(self):
        """Get the next protonated SMILES string.

        :raises StopIteration: If there are no more lines left iin the file.
        :return: TODO A dict, where the "smiles" key contains the canonical SMILES
                 string and the "data" key contains the remaining information
                 (e.g., the molecule name).
        :rtype: dict
        """

        # If there are any SMILES strings in self.cur_prot_SMI, just return
        # the first one and update the list to include only the remaining.
        if len(self.cur_prot_SMI) > 0:
            first, self.cur_prot_SMI = self.cur_prot_SMI[0], self.cur_prot_SMI[1:]
            return first

        # self.cur_prot_SMI is empty, so try to add more to it.

        # Get the next SMILES string from the input file.
        try:
            smile_and_datum = self.args["smiles_and_data"].next()
        except StopIteration:
            # There are no more input smiles strings...
            raise StopIteration()

        smi = smile_and_datum["smiles"]
        data = smile_and_datum["data"]  # Everything on SMILES line but the
                                        # SMILES string itself (e.g., the
                                        # molecule name).

        # Collect the data associated with this smiles (e.g., the molecule
        # name).
        tag = " ".join(data)

        # sites is a list of (atom index, "PROTONATED|DEPROTONATED|BOTH").
        # Note that the second entry indicates what state the site SHOULD be
        # in (not the one it IS in per the SMILES string). It's calculated
        # based on the probablistic distributions obtained during training.
        sites = ProtSubstructFuncs.get_prot_sites_and_target_states(smi, self.subs)

        new_smis = [smi]
        for site in sites:
            # Make a new smiles with the correct protonation state. Note that
            # new_smis is a growing list. This is how multiple protonation
            # sites are handled.

            # new_smis_to_perhaps_add = ProtSubstructFuncs.protonate_site(new_smis, site)
            new_smis = ProtSubstructFuncs.protonate_site(new_smis, site)
            # print(site, new_smis)  # Good for debugging.

            # Only add new smiles if not already in the list.
            # for s in new_smis_to_perhaps_add:
                # if not s in new_smis:
                    # new_smis.append(s)

        # In some cases, the script might generate redundant molecules.
        # Phosphonates, when the pH is between the two pKa values and the
        # stdev value is big enough, for example, will generate two identical
        # BOTH states. Let's remove this redundancy.
        new_smis = list(set(new_smis))

        # Deprotonating protonated aromatic nitrogen gives [nH-]. Change this
        # to [n-]. This is a hack.
        new_smis = [s.replace("[nH-]", "[n-]") for s in new_smis]

        # Sometimes Dimorphite-DL generates molecules that aren't actually
        # possible. Simply convert these to mol objects to eliminate the bad
        # ones (that are None).
        new_smis = [s for s in new_smis if UtilFuncs.convert_smiles_str_to_mol(s) is not None]

        # If there are no smi left, return the input one at the very least.
        # All generated forms have apparently been judged
        # inappropriate/mal-formed.
        if len(new_smis) == 0:
            new_smis = [smi]

        # If the user wants to see the target states, add those
        # to the ends of each line.
        if self.args["label_states"]:
            states = '\t'.join([x[1] for x in sites])
            new_lines = [x + "\t" + tag + "\t" + states for x in new_smis]
        else:
            new_lines = [x + "\t" + tag for x in new_smis]

        self.cur_prot_SMI = new_lines

        return self.next()

class ProtSubstructFuncs:
    """A namespace to store functions for loading the substructures that can
    be protonated. To keep things organized."""

    @staticmethod
    def load_protonation_substructs_calc_state_for_ph(min_ph=6.4, max_ph=8.4, pka_std_range=1):
        """A pre-calculated list of R-groups with protonation sites, with their
        likely pKa bins.

        :param float min_ph:  The lower bound on the pH range, defaults to 6.4.
        :param float max_ph:  The upper bound on the pH range, defaults to 8.4.
        :param pka_std_range: Basically the precision (stdev from predicted pKa to
                            consider), defaults to 1.
        :return: A dict of the protonation substructions for the specified pH
                range.
        """

        subs = []
        pwd = os.path.dirname(os.path.realpath(__file__))

        site_structures_file = "{}/{}".format(pwd, "site_substructures.smarts")
        with open(site_structures_file, 'r') as substruct:
            for line in substruct:
                line = line.strip()
                sub = {}
                if line is not "":
                    splits = line.split()
                    sub["name"] = splits[0]
                    sub["smart"] = splits[1]
                    sub["mol"] = Chem.MolFromSmarts(sub["smart"])

                    # NEED TO DIVIDE THIS BY 3s
                    pka_ranges = [splits[i:i+3] for i in range(2, len(splits)-1, 3)]

                    prot = []
                    for pka_range in pka_ranges:
                        site = pka_range[0]
                        std = float(pka_range[2]) * pka_std_range
                        mean = float(pka_range[1])
                        protonation_state = ProtSubstructFuncs.define_protonation_state(
                            mean, std, min_ph, max_ph
                        )

                        prot.append([site, protonation_state])

                    sub["prot_states_for_pH"] = prot
                    subs.append(sub)
        return subs

    @staticmethod
    def define_protonation_state(mean, std, min_ph, max_ph):
        """Updates the substructure definitions to include the protonation state
        based on the user-given pH range. The size of the pKa range is also based
        on the number of standard deviations to be considered by the user param.

        :param float mean:   The mean pKa.
        :param float std:    The precision (stdev).
        :param float min_ph: The min pH of the range.
        :param float max_ph: The max pH of the range.
        :return: A string describing the protonation state.
        """

        min_pka = mean - std
        max_pka = mean + std

        # This needs to be reassigned, and 'ERROR' should never make it past the
        # next set of checks.
        if min_pka <= max_ph and min_ph <= max_pka:
            protonation_state = 'BOTH'
        elif mean > max_ph:
            protonation_state = 'PROTONATED'
        else:
            protonation_state = 'DEPROTONATED'

        return protonation_state

    @staticmethod
    def get_prot_sites_and_target_states(smi, subs):
        """For a single molecule, find all possible matches in the protonation
        R-group list, subs. Items that are higher on the list will be matched
        first, to the exclusion of later items.

        :param string smi: A SMILES string.
        :param list subs: Substructure information.
        :return: A list of protonation sites and their pKa bin. ('PROTONATED',
            'BOTH', or  'DEPROTONATED')
        """

        # Convert the Smiles string (smi) to an RDKit Mol Obj
        mol = UtilFuncs.convert_smiles_str_to_mol(smi)

        # Check Conversion worked
        if mol is None:
            UtilFuncs.eprint("ERROR:   ", smi)
            return []

        # Try to Add hydrogens. if failed return []
        try:
            mol =  Chem.AddHs(mol)
        except:
            UtilFuncs.eprint("ERROR:   ", smi)
            return []

        # Check adding Hs worked
        if mol is None:
            UtilFuncs.eprint("ERROR:   ", smi)
            return []

        ProtectUnprotectFuncs.unprotect_molecule(mol)
        protonation_sites = []

        for item in subs:
            smart = item["mol"]
            if mol.HasSubstructMatch(smart):
                matches = ProtectUnprotectFuncs.get_unprotected_matches(mol, smart)
                prot = item["prot_states_for_pH"]
                for match in matches:
                    # We want to move the site from being relative to the
                    # substructure, to the index on the main molecule.
                    for site in prot:
                        proton = int(site[0])
                        category = site[1]
                        new_site = (match[proton], category, item["name"])

                        if not new_site in protonation_sites:
                            # Because sites must be unique.
                            protonation_sites.append(new_site)

                    ProtectUnprotectFuncs.protect_molecule(mol, match)

        return protonation_sites

    @staticmethod
    def protonate_site(smis, site):
        """Given a list of SMILES strings, we protonate the site.

        :param list smis:  The list of SMILES strings.
        :param tuple site: Information about the protonation site.
                        (idx, target_prot_state, prot_site_name)
        :return: A list of the appropriately protonated SMILES.
        """

        # Decouple the atom index and its target protonation state from the site
        # tuple
        idx, target_prot_state, prot_site_name = site

        # Initialize the output list
        output_smis = []

        state_to_charge = {"DEPROTONATED": [-1],
                        "PROTONATED": [0],
                        "BOTH": [-1, 0]}

        charges = state_to_charge[target_prot_state]

        # Now make the actual smiles match the target protonation state.
        output_smis = ProtSubstructFuncs.set_protonation_charge(smis, idx, charges, prot_site_name)

        return output_smis

    @staticmethod
    def set_protonation_charge(smis, idx, charges, prot_site_name):
        """Sets the atomic charge on a particular site for a set of SMILES.

        :param list smis:             A list of the SMILES strings.
        :param int idx:               The index of the atom to consider.
        :param list charges:          A list of the charges (ints) to assign at
                                    this site.
        :param string prot_site_name: The name of the protonation site.
        :return: A list of the processed SMILES strings.
        """

        # Sets up the output list and the Nitrogen charge
        output = []

        for charge in charges:
            # The charge for Nitrogens is 1 higher than others (i.e., protonated
            # state is positively charged).
            nitro_charge = charge + 1

            # But there are a few nitrogen moieties where the acidic group is the
            # neutral one. Amides are a good example. I gave some thought re. how
            # to best flag these. I decided that those nitrogen-containing
            # moieties where the acidic group is neutral (rather than positively
            # charged) will have "*" in the name.
            if "*" in prot_site_name:
                nitro_charge = nitro_charge - 1  # Undo what was done previously.

            for smi in smis:

                # Convert smilesstring (smi) into a RDKit Mol
                mol = UtilFuncs.convert_smiles_str_to_mol(smi)

                # Check that the conversion worked, skip if it fails
                if mol is None:
                    continue

                atom = mol.GetAtomWithIdx(idx)

                # Assign the protonation charge, with special care for Nitrogens
                element = atom.GetAtomicNum()
                if element == 7:
                    atom.SetFormalCharge(nitro_charge)
                else:
                    atom.SetFormalCharge(charge)

                # Convert back to SMILE and add to output
                out_smile = Chem.MolToSmiles(mol, isomericSmiles=True,canonical=True)
                output.append(out_smile)

        return output

class ProtectUnprotectFuncs:
    """A namespace for storing functions that are useful for protecting and
    unprotecting molecules. To keep things organized. We need to identify and
    mark groups that have been matched with a substructure."""

    @staticmethod
    def unprotect_molecule(mol):
        """Sets the protected property on all atoms to 0. This also creates the
        property for new molecules.

        :param rdkit.Chem.rdchem.Mol mol: The rdkit Mol object.
        :type mol: The rdkit Mol object with atoms unprotected.
        """

        for atom in mol.GetAtoms():
            atom.SetProp('_protected', '0')

    @staticmethod
    def protect_molecule(mol, match):
        """Given a 'match', a list of molecules idx's, we set the protected status
        of each atom to 1. This will prevent any matches using that atom in the
        future.

        :param rdkit.Chem.rdchem.Mol mol: The rdkit Mol object to protect.
        :param list match: A list of molecule idx's.
        """

        for idx in match:
            atom = mol.GetAtomWithIdx(idx)
            atom.SetProp('_protected', '1')

    @staticmethod
    def get_unprotected_matches(mol, substruct):
        """Finds substructure matches with atoms that have not been protected.
        Returns list of matches, each match a list of atom idxs.

        :param rdkit.Chem.rdchem.Mol mol: The Mol object to consider.
        :param string substruct: The SMARTS string of the substructure ot match.
        :return: A list of the matches. Each match is itself a list of atom idxs.
        """

        matches = mol.GetSubstructMatches(substruct)
        unprotected_matches = []
        for match in matches:
            if ProtectUnprotectFuncs.is_match_unprotected(mol, match):
                unprotected_matches.append(match)
        return unprotected_matches

    @staticmethod
    def is_match_unprotected(mol, match):
        """Checks a molecule to see if the substructure match contains any
        protected atoms.

        :param rdkit.Chem.rdchem.Mol mol: The Mol object to check.
        :param list match: The match to check.
        :return: A boolean, whether the match is present or not.
        """

        for idx in match:
            atom = mol.GetAtomWithIdx(idx)
            protected = atom.GetProp("_protected")
            if protected == "1":
                return False
        return True

class TestFuncs:
    """A namespace for storing functions that perform tests on the code. To
    keep things organized."""

    @staticmethod
    def test():
        """Tests all the 38 groups."""

        smis = [
            # [input smiles, pka, protonated, deprotonated, category]
            ["C#CCO",                  "C#CCO",                     "C#CC[O-]",                 "Alcohol"],
            ["C(=O)N",                 "NC=O",                      "[NH-]C=O",                 "Amide"],
            ["CC(=O)NOC(C)=O",         "CC(=O)NOC(C)=O",            "CC(=O)[N-]OC(C)=O",        "Amide_electronegative"],
            ["COC(=N)N",               "COC(N)=[NH2+]",             "COC(=N)N",                 "AmidineGuanidine2"],
            ["Brc1ccc(C2NCCS2)cc1",    "Brc1ccc(C2[NH2+]CCS2)cc1",  "Brc1ccc(C2NCCS2)cc1",      "Amines_primary_secondary_tertiary"],
            ["CC(=O)[n+]1ccc(N)cc1",   "CC(=O)[n+]1ccc([NH3+])cc1", "CC(=O)[n+]1ccc(N)cc1",     "Anilines_primary"],
            ["CCNc1ccccc1",            "CC[NH2+]c1ccccc1",          "CCNc1ccccc1",              "Anilines_secondary"],
            ["Cc1ccccc1N(C)C",         "Cc1ccccc1[NH+](C)C",        "Cc1ccccc1N(C)C",           "Anilines_tertiary"],
            ["BrC1=CC2=C(C=C1)NC=C2",  "Brc1ccc2[nH]ccc2c1",        "Brc1ccc2[n-]ccc2c1",       "Indole_pyrrole"],
            ["O=c1cc[nH]cc1",          "O=c1cc[nH]cc1",             "O=c1cc[n-]cc1",            "Aromatic_nitrogen_protonated"],
            ["C-N=[N+]=[N@H]",         "CN=[N+]=N",                 "CN=[N+]=[N-]",             "Azide"],
            ["BrC(C(O)=O)CBr",         "O=C(O)C(Br)CBr",            "O=C([O-])C(Br)CBr",        "Carboxyl"],
            ["NC(NN=O)=N",             "NC(=[NH2+])NN=O",           "N=C(N)NN=O",               "AmidineGuanidine1"],
            ["C(F)(F)(F)C(=O)NC(=O)C", "CC(=O)NC(=O)C(F)(F)F",      "CC(=O)[N-]C(=O)C(F)(F)F",  "Imide"],
            ["O=C(C)NC(C)=O",          "CC(=O)NC(C)=O",             "CC(=O)[N-]C(C)=O",         "Imide2"],
            ["CC(C)(C)C(N(C)O)=O",     "CN(O)C(=O)C(C)(C)C",        "CN([O-])C(=O)C(C)(C)C",    "N-hydroxyamide"],
            ["C[N+](O)=O",             "C[N+](=O)O",                "C[N+](=O)[O-]",            "Nitro"],
            ["O=C1C=C(O)CC1",          "O=C1C=C(O)CC1",             "O=C1C=C([O-])CC1",         "O=C-C=C-OH"],
            ["C1CC1OO",                "OOC1CC1",                   "[O-]OC1CC1",               "Peroxide2"],
            ["C(=O)OO",                "O=COO",                     "O=CO[O-]",                 "Peroxide1"],
            ["Brc1cc(O)cc(Br)c1",      "Oc1cc(Br)cc(Br)c1",         "[O-]c1cc(Br)cc(Br)c1",     "Phenol"],
            ["CC(=O)c1ccc(S)cc1",      "CC(=O)c1ccc(S)cc1",         "CC(=O)c1ccc([S-])cc1",     "Phenyl_Thiol"],
            ["C=CCOc1ccc(C(=O)O)cc1",  "C=CCOc1ccc(C(=O)O)cc1",     "C=CCOc1ccc(C(=O)[O-])cc1", "Phenyl_carboxyl"],
            ["COP(=O)(O)OC",           "COP(=O)(O)OC",              "COP(=O)([O-])OC",          "Phosphate_diester"],
            ["CP(C)(=O)O",             "CP(C)(=O)O",                "CP(C)(=O)[O-]",            "Phosphinic_acid"],
            ["CC(C)OP(C)(=O)O",        "CC(C)OP(C)(=O)O",           "CC(C)OP(C)(=O)[O-]",       "Phosphonate_ester"],
            ["CC1(C)OC(=O)NC1=O",      "CC1(C)OC(=O)NC1=O",         "CC1(C)OC(=O)[N-]C1=O",     "Ringed_imide1"],
            ["O=C(N1)C=CC1=O",         "O=C1C=CC(=O)N1",            "O=C1C=CC(=O)[N-]1",        "Ringed_imide2"],
            ["O=S(OC)(O)=O",           "COS(=O)(=O)O",              "COS(=O)(=O)[O-]",          "Sulfate"],
            ["COc1ccc(S(=O)O)cc1",     "COc1ccc(S(=O)O)cc1",        "COc1ccc(S(=O)[O-])cc1",    "Sulfinic_acid"],
            ["CS(N)(=O)=O",            "CS(N)(=O)=O",               "CS([NH-])(=O)=O",          "Sulfonamide"],
            ["CC(=O)CSCCS(O)(=O)=O",   "CC(=O)CSCCS(=O)(=O)O",      "CC(=O)CSCCS(=O)(=O)[O-]",  "Sulfonate"],
            ["CC(=O)S",                "CC(=O)S",                   "CC(=O)[S-]",               "Thioic_acid"],
            ["C(C)(C)(C)(S)",          "CC(C)(C)S",                 "CC(C)(C)[S-]",             "Thiol"],
            ["Brc1cc[nH+]cc1",         "Brc1cc[nH+]cc1",            "Brc1ccncc1",               "Aromatic_nitrogen_unprotonated"],
            ["C=C(O)c1c(C)cc(C)cc1C",  "C=C(O)c1c(C)cc(C)cc1C",     "C=C([O-])c1c(C)cc(C)cc1C", "Vinyl_alcohol"],
            ["CC(=O)ON",               "CC(=O)O[NH3+]",             "CC(=O)ON",                 "Primary_hydroxyl_amine"]
        ]

        smis_phos = [
            ["O=P(O)(O)OCCCC", "CCCCOP(=O)(O)O", "CCCCOP(=O)([O-])O", "CCCCOP(=O)([O-])[O-]", "Phosphate"],
            ["CC(P(O)(O)=O)C", "CC(C)P(=O)(O)O", "CC(C)P(=O)([O-])O", "CC(C)P(=O)([O-])[O-]", "Phosphonate"]
        ]

        # Load the average pKa values.
        average_pkas = {l.split()[0].replace("*", ""):float(l.split()[3]) for l in open("site_substructures.smarts") if l.split()[0] not in ["Phosphate", "Phosphonate"]}
        average_pkas_phos = {l.split()[0].replace("*", ""):[float(l.split()[3]), float(l.split()[6])] for l in open("site_substructures.smarts") if l.split()[0] in ["Phosphate", "Phosphonate"]}

        print("Running Tests")
        print("=============")
        print("")

        print("Very Acidic (pH -10000000)")
        print("--------------------------")
        print("")

        args = {
            "min_ph": -10000000,
            "max_ph": -10000000,
            "pka_precision": 0.5,
            "smiles": "",
            "label_states": True
        }

        for smi, protonated, deprotonated, category in smis:
            args["smiles"] = smi
            TestFuncs.test_check(args, [protonated], ["PROTONATED"])

        for smi, protonated, mix, deprotonated, category in smis_phos:
            args["smiles"] = smi
            TestFuncs.test_check(args, [protonated], ["PROTONATED"])

        args["min_ph"] = 10000000
        args["max_ph"] = 10000000

        print("")
        print("Very Basic (pH 10000000)")
        print("------------------------")
        print("")

        for smi, protonated, deprotonated, category in smis:
            args["smiles"] = smi
            TestFuncs.test_check(args, [deprotonated], ["DEPROTONATED"])

        for smi, protonated, mix, deprotonated, category in smis_phos:
            args["smiles"] = smi
            TestFuncs.test_check(args, [deprotonated], ["DEPROTONATED"])

        print("")
        print("pH is Category pKa")
        print("------------------")
        print("")

        for smi, protonated, deprotonated, category in smis:
            avg_pka = average_pkas[category]

            args["smiles"] = smi
            args["min_ph"] = avg_pka
            args["max_ph"] = avg_pka

            TestFuncs.test_check(args, [protonated, deprotonated], ["BOTH"])

        for smi, protonated, mix, deprotonated, category in smis_phos:
            args["smiles"] = smi

            avg_pka = average_pkas_phos[category][0]
            args["min_ph"] = avg_pka
            args["max_ph"] = avg_pka

            TestFuncs.test_check(args, [mix, protonated], ["BOTH"])

            avg_pka = average_pkas_phos[category][1]
            args["min_ph"] = avg_pka
            args["max_ph"] = avg_pka

            TestFuncs.test_check(args, [mix, deprotonated], ["DEPROTONATED", "DEPROTONATED"])

            avg_pka = 0.5 * (average_pkas_phos[category][0] + average_pkas_phos[category][1])
            args["min_ph"] = avg_pka
            args["max_ph"] = avg_pka
            args["pka_precision"] = 5  # Should give all three

            TestFuncs.test_check(args, [mix, deprotonated, protonated], ["BOTH", "BOTH"])

    @staticmethod
    def test_check(args, expected_output, labels):
        """Tests most ionizable groups. The ones that can only loose or gain a single proton.

        :param args: The arguments to pass to protonate()
        :param expected_output: A list of the expected SMILES-strings output.
        :param labels: The labels. A list containing combo of BOTH, PROTONATED,
                    DEPROTONATED.
        :raises Exception: Wrong number of states produced.
        :raises Exception: Unexpected output SMILES.
        :raises Exception: Wrong labels.
        """

        output = list(Protonate(args))
        output = [o.split() for o in output]

        num_states = len(expected_output)

        if (len(output) != num_states):
            msg = args["smiles"] + " should have " + str(num_states) + \
                " states at at pH " + str(args["min_ph"]) + ": " + str(output)
            print(msg)
            raise Exception(msg)

        if (len(set([l[0] for l in output]) - set(expected_output)) != 0):
            msg = args["smiles"] + " is not " + " AND ".join(expected_output) + \
                " at pH " + str(args["min_ph"]) + " - " + str(args["max_ph"]) + \
                "; it is " + " AND ".join([l[0] for l in output])
            print(msg)
            raise Exception(msg)

        if (len(set([l[1] for l in output]) - set(labels)) != 0):
            msg = args["smiles"] + " not labeled as " + " AND ".join(labels) + \
                "; it is " + " AND ".join([l[1] for l in output])
            print(msg)
            raise Exception(msg)

        ph_range = sorted(list(set([args["min_ph"], args["max_ph"]])))
        ph_range_str = "(" + " - ".join("{0:.2f}".format(n) for n in ph_range) + ")"
        print("(CORRECT) " + ph_range_str.ljust(10) + " " + args["smiles"] + " => " + " AND ".join([l[0] for l in output]))

def run(**kwargs):
    """A helpful, importable function for those who want to call Dimorphite-DL
    from another Python script rather than the command line. Note that this
    function accepts keyword arguments that match the command-line parameters
    exactly. If you want to pass and return a list of RDKit Mol objects, import
    run_with_mol_list() instead.

    :param **kwargs: For a complete description, run dimorphite_dl.py from the
        command line with the -h option.
    :type kwargs: dict
    """

    # Run the main function with the specified arguments.
    main(kwargs)

def run_with_mol_list(mol_lst, **kwargs):
    """A helpful, importable function for those who want to call Dimorphite-DL
    from another Python script rather than the command line. Note that this
    function is for passing Dimorphite-DL a list of RDKit Mol objects, together
    with command-line parameters. If you want to use only the same parameters
    that you would use from the command line, import run() instead.

    :param mol_lst: A list of rdkit.Chem.rdchem.Mol objects.
    :type mol_lst: list
    :raises Exception: If the **kwargs includes "smiles", "smiles_file",
                       "output_file", or "test" parameters.
    :return: A list of properly protonated rdkit.Chem.rdchem.Mol objects.
    :rtype: list
    """

    # Do a quick check to make sure the user input makes sense.
    for bad_arg in ["smiles", "smiles_file", "output_file", "test"]:
        if bad_arg in kwargs:
            msg = "You're using Dimorphite-DL's run_with_mol_list(mol_lst, " + \
                   "**kwargs) function, but you also passed the \"" + \
                   bad_arg + "\" argument. Did you mean to use the " + \
                   "run(**kwargs) function instead?"
            print(msg)
            raise Exception(msg)

    # Set the return_as_list flag so main() will return the protonated smiles
    # as a list.
    kwargs["return_as_list"] = True

    # Having reviewed the code, it will be very difficult to rewrite it so
    # that a list of Mol objects can be used directly. Intead, convert this
    # list of mols to smiles and pass that. Not efficient, but it will work.
    protonated_smiles_and_props = []
    for m in mol_lst:
        props = m.GetPropsAsDict()
        kwargs["smiles"] = Chem.MolToSmiles(m, isomericSmiles=True)
        protonated_smiles_and_props.extend(
            [(s.split("\t")[0], props) for s in main(kwargs)]
        )

    # Now convert the list of protonated smiles strings back to RDKit Mol
    # objects. Also, add back in the properties from the original mol objects.
    mols = []
    for s, props in protonated_smiles_and_props:
        m = Chem.MolFromSmiles(s)
        if m:
            for prop, val in props.items():
                if type(val) is int:
                    m.SetIntProp(prop, val)
                elif type(val) is float:
                    m.SetDoubleProp(prop, val)
                elif type(val) is bool:
                    m.SetBoolProp(prop, val)
                else:
                    m.SetProp(prop, str(val))
            mols.append(m)
        else:
            UtilFuncs.eprint("WARNING: Could not process molecule with SMILES string " + s + " and properties " + str(props))

    return mols

if __name__ == "__main__":
    main()