File size: 26,122 Bytes
233f6d4
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
import logging, sys

import numpy as np
from tqdm import tqdm
from scipy.linalg import eigh

from rdkit import Chem
from rdkit.Chem import MolFromSmiles, MolToSmiles, AddHs

from nfp.preprocessing import features
from nfp.preprocessing.features import Tokenizer
import time


class SmilesPreprocessor(object):
    """ Given a list of SMILES strings, encode these molecules as atom and
    connectivity feature matricies.

    Example:
    >>> preprocessor = SmilesPreprocessor(explicit_hs=False)
    >>> inputs = preprocessor.fit(data.smiles)
    """

    def __init__(self, explicit_hs=True, atom_features=None, bond_features=None):
        """

        explicit_hs : bool
            whether to tell RDkit to add H's to a molecule.
        atom_features : function
            A function applied to an rdkit.Atom that returns some
            representation (i.e., string, integer) for the Tokenizer class.
        bond_features : function
            A function applied to an rdkit Bond to return some description.

        """

        self.atom_tokenizer = Tokenizer()
        self.bond_tokenizer = Tokenizer()
        self.explicit_hs = explicit_hs

        if atom_features is None:
            atom_features = features.atom_features_v1

        if bond_features is None:
            bond_features = features.bond_features_v1

        self.atom_features = atom_features
        self.bond_features = bond_features


    def fit(self, smiles_iterator):
        """ Fit an iterator of SMILES strings, creating new atom and bond
        tokens for unseen molecules. Returns a dictionary with 'atom' and
        'connectivity' entries """
        return list(self.preprocess(smiles_iterator, train=True))


    def predict(self, smiles_iterator):
        """ Uses previously determined atom and bond tokens to convert a SMILES
        iterator into 'atom' and 'connectivity' matrices. Ensures that atom and
        bond classes commute with previously determined results. """
        return list(self.preprocess(smiles_iterator, train=False))


    def preprocess(self, smiles_iterator, train=True):

        self.atom_tokenizer.train = train
        self.bond_tokenizer.train = train

        for smiles in tqdm(smiles_iterator):
            yield self.construct_feature_matrices(smiles)


    @property
    def atom_classes(self):
        """ The number of atom types found (includes the 0 null-atom type) """
        return self.atom_tokenizer.num_classes + 1


    @property
    def bond_classes(self):
        """ The number of bond types found (includes the 0 null-bond type) """
        return self.bond_tokenizer.num_classes + 1


    def construct_feature_matrices(self, smiles):
        """ construct a molecule from the given smiles string and return atom
        and bond classes.

        Returns
        dict with entries
        'n_atom' : number of atoms in the molecule
        'n_bond' : number of bonds in the molecule 
        'atom' : (n_atom,) length list of atom classes
        'bond' : (n_bond,) list of bond classes
        'connectivity' : (n_bond, 2) array of source atom, target atom pairs.

        """

        mol = MolFromSmiles(smiles)
        if self.explicit_hs:
            mol = AddHs(mol)

        n_atom = len(mol.GetAtoms())
        n_bond = 2 * len(mol.GetBonds())

        # If its an isolated atom, add a self-link
        if n_bond == 0:
            n_bond = 1
        
        atom_feature_matrix = np.zeros(n_atom, dtype='int')
        bond_feature_matrix = np.zeros(n_bond, dtype='int')
        connectivity = np.zeros((n_bond, 2), dtype='int')

        bond_index = 0

        atom_seq = mol.GetAtoms()
        atoms = [atom_seq[i] for i in range(n_atom)]

        for n, atom in enumerate(atoms):

            # Atom Classes
            atom_feature_matrix[n] = self.atom_tokenizer(
                self.atom_features(atom))

            start_index = atom.GetIdx()

            for bond in atom.GetBonds():
                # Is the bond pointing at the target atom
                rev = bond.GetBeginAtomIdx() != start_index

                # Bond Classes
                bond_feature_matrix[n] = self.bond_tokenizer(
                    self.bond_features(bond, flipped=rev))

                # Connectivity
                if not rev:  # Original direction
                    connectivity[bond_index, 0] = bond.GetBeginAtomIdx()
                    connectivity[bond_index, 1] = bond.GetEndAtomIdx()

                else:  # Reversed
                    connectivity[bond_index, 0] = bond.GetEndAtomIdx()
                    connectivity[bond_index, 1] = bond.GetBeginAtomIdx()

                bond_index += 1


        return {
            'n_atom': n_atom,
            'n_bond': n_bond,
            'atom': atom_feature_matrix,
            'bond': bond_feature_matrix,
            'connectivity': connectivity,
        }
    

class ConnectivityAPreprocessor(object):
    """ Given a list of SMILES strings, encode these molecules as atom and
    connectivity feature matricies.

    Example:
    >>> preprocessor = SmilesPreprocessor(explicit_hs=False)
    >>> inputs = preprocessor.fit(data.smiles)
    """

    def __init__(self, explicit_hs=True, atom_features=None, bond_features=None):
        """

        explicit_hs : bool
            whether to tell RDkit to add H's to a molecule.
        atom_features : function
            A function applied to an rdkit.Atom that returns some
            representation (i.e., string, integer) for the Tokenizer class.
        bond_features : function
            A function applied to an rdkit Bond to return some description.

        """

        self.atom_tokenizer = Tokenizer()
        self.bond_tokenizer = Tokenizer()
        self.explicit_hs = explicit_hs

        if atom_features is None:
            atom_features = features.atom_features_v1

        if bond_features is None:
            bond_features = features.bond_features_v1

        self.atom_features = atom_features
        self.bond_features = bond_features


    def fit(self, smiles_iterator):
        """ Fit an iterator of SMILES strings, creating new atom and bond
        tokens for unseen molecules. Returns a dictionary with 'atom' and
        'connectivity' entries """
        return list(self.preprocess(smiles_iterator, train=True))


    def predict(self, smiles_iterator):
        """ Uses previously determined atom and bond tokens to convert a SMILES
        iterator into 'atom' and 'connectivity' matrices. Ensures that atom and
        bond classes commute with previously determined results. """
        return list(self.preprocess(smiles_iterator, train=False))


    def preprocess(self, smiles_iterator, train=True):

        self.atom_tokenizer.train = train
        self.bond_tokenizer.train = train

        for smiles in tqdm(smiles_iterator):
            yield self.construct_feature_matrices(smiles)


    @property
    def atom_classes(self):
        """ The number of atom types found (includes the 0 null-atom type) """
        return self.atom_tokenizer.num_classes + 1


    @property
    def bond_classes(self):
        """ The number of bond types found (includes the 0 null-bond type) """
        return self.bond_tokenizer.num_classes + 1


    def construct_feature_matrices(self, smiles):
        """ construct a molecule from the given smiles string and return atom
        and bond classes.

        Returns
        dict with entries
        'n_atom' : number of atoms in the molecule
        'n_bond' : number of bonds in the molecule 
        'atom' : (n_atom,) length list of atom classes
        'bond' : (n_bond,) list of bond classes
        'connectivity' : (n_bond, 2) array of source atom, target atom pairs.

        """

        mol = MolFromSmiles(smiles)
        if self.explicit_hs:
            mol = AddHs(mol)

        n_atom = len(mol.GetAtoms())
        n_bond = 2 * len(mol.GetBonds())

        # If its an isolated atom, add a self-link
        if n_bond == 0:
            n_bond = 1
        
        atom_feature_matrix = np.zeros(n_atom, dtype='int')
        bond_feature_matrix = np.zeros(n_bond, dtype='int')
        connectivity = np.zeros((n_bond, 2), dtype='int')

        bond_index = 0

        atom_seq = mol.GetAtoms()
        atoms = [atom_seq[i] for i in range(n_atom)]

        for n, atom in enumerate(atoms):

            # Atom Classes
            atom_feature_matrix[n] = self.atom_tokenizer(
                self.atom_features(atom))

            start_index = atom.GetIdx()

            for bond in atom.GetBonds():
                # Is the bond pointing at the target atom
                rev = bond.GetBeginAtomIdx() != start_index

                # Bond Classes
                bond_feature_matrix[n] = self.bond_tokenizer(
                    self.bond_features(bond, flipped=rev))

                # Connectivity
                if not rev:  # Original direction
                    connectivity[bond_index, 0] = bond.GetBeginAtomIdx()
                    connectivity[bond_index, 1] = bond.GetEndAtomIdx()

                else:  # Reversed
                    connectivity[bond_index, 0] = bond.GetEndAtomIdx()
                    connectivity[bond_index, 1] = bond.GetBeginAtomIdx()

                bond_index += 1


        return {
            'n_atom': n_atom,
            'n_bond': n_bond,
            'atom': atom_feature_matrix,
            'bond': bond_feature_matrix,
            'connectivity': connectivity,
        }


class MolPreprocessor(SmilesPreprocessor):
    """ I should refactor this into a base class and separate
    SmilesPreprocessor classes. But the idea is that we only need to redefine
    the `construct_feature_matrices` method to have a working preprocessor that
    handles 3D structures. 

    We'll pass an iterator of mol objects instead of SMILES strings this time,
    though.
    
    """

    def __init__(self, n_neighbors, cutoff, **kwargs):
        """ A preprocessor class that also returns distances between
        neighboring atoms. Adds edges for non-bonded atoms to include a maximum
        of n_neighbors around each atom """

        self.n_neighbors = n_neighbors
        self.cutoff = cutoff
        super(MolPreprocessor, self).__init__(**kwargs)


    def construct_feature_matrices(self, mol):
        """ Given an rdkit mol, return atom feature matrices, bond feature
        matrices, and connectivity matrices.

        Returns
        dict with entries
        'n_atom' : number of atoms in the molecule
        'n_bond' : number of edges (likely n_atom * n_neighbors)
        'atom' : (n_atom,) length list of atom classes
        'bond' : (n_bond,) list of bond classes. 0 for no bond
        'distance' : (n_bond,) list of bond distances
        'connectivity' : (n_bond, 2) array of source atom, target atom pairs.
            
        """

        n_atom = len(mol.GetAtoms())

        # n_bond is actually the number of atom-atom pairs, so this is defined
        # by the number of neighbors for each atom.
        #if there is cutoff, 
        distance_matrix = Chem.Get3DDistanceMatrix(mol)

        if self.n_neighbors <= (n_atom - 1):
            n_bond = self.n_neighbors * n_atom
        else:
            # If there are fewer atoms than n_neighbors, all atoms will be
            # connected
            n_bond = distance_matrix[(distance_matrix < self.cutoff) & (distance_matrix != 0)].size

        if n_bond == 0: n_bond = 1

        # Initialize the matrices to be filled in during the following loop.
        atom_feature_matrix = np.zeros(n_atom, dtype='int')
        bond_feature_matrix = np.zeros(n_bond, dtype='int')
        bond_distance_matrix = np.zeros(n_bond, dtype=np.float32)
        connectivity = np.zeros((n_bond, 2), dtype='int')

        # Hopefully we've filtered out all problem mols by now.
        if mol is None:
            raise RuntimeError("Issue in loading mol")
        
        # Get a list of the atoms in the molecule.
        atom_seq = mol.GetAtoms()
        atoms = [atom_seq[i] for i in range(n_atom)]

        # Here we loop over each atom, and the inner loop iterates over each
        # neighbor of the current atom.
        bond_index = 0  # keep track of our current bond.
        for n, atom in enumerate(atoms):
            
            # update atom feature matrix
            atom_feature_matrix[n] = self.atom_tokenizer(
                self.atom_features(atom))
            
            # if n_neighbors is greater than total atoms, then each atom is a
            # neighbor.
            if (self.n_neighbors + 1) > len(mol.GetAtoms()):
                neighbor_end_index = len(mol.GetAtoms())
            else:
                neighbor_end_index = (self.n_neighbors + 1)

            distance_atom = distance_matrix[n, :]
            cutoff_end_index = distance_atom[distance_atom < self.cutoff].size

            end_index = min(neighbor_end_index, cutoff_end_index)

            # Loop over each of the nearest neighbors

            neighbor_inds = distance_matrix[n, :].argsort()[1:end_index]
            if len(neighbor_inds)==0: neighbor_inds = [n]
            for neighbor in neighbor_inds:
                
                # update bond feature matrix
                bond = mol.GetBondBetweenAtoms(n, int(neighbor))
                if bond is None:
                    bond_feature_matrix[bond_index] = 0
                else:
                    rev = False if bond.GetBeginAtomIdx() == n else True
                    bond_feature_matrix[bond_index] = self.bond_tokenizer(
                        self.bond_features(bond, flipped=rev))

                distance = distance_matrix[n, neighbor]
                bond_distance_matrix[bond_index] = distance
                
                # update connectivity matrix
                connectivity[bond_index, 0] = n
                connectivity[bond_index, 1] = neighbor
                
                bond_index += 1

        return {
            'n_atom': n_atom,
            'n_bond': n_bond,
            'atom': atom_feature_matrix,
            'bond': bond_feature_matrix,
            'distance': bond_distance_matrix,
            'connectivity': connectivity,
        }


class MolBPreprocessor(MolPreprocessor):
    """
    This is a subclass of Molpreprocessor that preprocessor molecule with
    bond property target
    """
    def __init__(self, **kwargs):
        """
        A preprocessor class that also returns bond_target_matrix, besides the bond matrix
        returned by MolPreprocessor. The bond_target_matrix is then used as ref to reduce molecule
        to bond property
        """
        super(MolBPreprocessor, self).__init__(**kwargs)

    def construct_feature_matrices(self, entry):
        """
        Given an entry contining rdkit molecule, bond_index and for the target property, 
        return atom 
        feature matrices, bond feature matrices, distance matrices, connectivity matrices and bond
        ref matrices.

        returns
        dict with entries
        see MolPreproccessor
        'bond_index' : ref array to the bond index
        """
        mol, bond_index_array = entry
        
        n_atom = len(mol.GetAtoms())
        n_pro = len(bond_index_array)

        # n_bond is actually the number of atom-atom pairs, so this is defined
        # by the number of neighbors for each atom.
        #if there is cutoff, 
        distance_matrix = Chem.Get3DDistanceMatrix(mol)

        if self.n_neighbors <= (n_atom - 1):
            n_bond = self.n_neighbors * n_atom
        else:
            # If there are fewer atoms than n_neighbors, all atoms will be
            # connected
            n_bond = distance_matrix[(distance_matrix < self.cutoff) & (distance_matrix != 0)].size

        if n_bond == 0: n_bond = 1

        # Initialize the matrices to be filled in during the following loop.
        atom_feature_matrix = np.zeros(n_atom, dtype='int')
        bond_feature_matrix = np.zeros(n_bond, dtype='int')
        bond_distance_matrix = np.zeros(n_bond, dtype=np.float32)
        bond_index_matrix = np.full(n_bond, -1, dtype='int')
        connectivity = np.zeros((n_bond, 2), dtype='int')

        # Hopefully we've filtered out all problem mols by now.
        if mol is None:
            raise RuntimeError("Issue in loading mol")
        
        # Get a list of the atoms in the molecule.
        atom_seq = mol.GetAtoms()
        atoms = [atom_seq[i] for i in range(n_atom)]

        # Here we loop over each atom, and the inner loop iterates over each
        # neighbor of the current atom.
        bond_index = 0  # keep track of our current bond.
        for n, atom in enumerate(atoms):
            # update atom feature matrix
            atom_feature_matrix[n] = self.atom_tokenizer(
                self.atom_features(atom))
            
            # if n_neighbors is greater than total atoms, then each atom is a
            # neighbor.
            if (self.n_neighbors + 1) > len(mol.GetAtoms()):
                neighbor_end_index = len(mol.GetAtoms())
            else:
                neighbor_end_index = (self.n_neighbors + 1)

            distance_atom = distance_matrix[n, :]
            cutoff_end_index = distance_atom[distance_atom < self.cutoff].size

            end_index = min(neighbor_end_index, cutoff_end_index)

            # Loop over each of the nearest neighbors

            neighbor_inds = distance_matrix[n, :].argsort()[1:end_index]
            if len(neighbor_inds)==0: neighbor_inds = [n]
            for neighbor in neighbor_inds:
                
                # update bond feature matrix
                bond = mol.GetBondBetweenAtoms(n, int(neighbor))
                if bond is None:
                    bond_feature_matrix[bond_index] = 0
                else:
                    rev = False if bond.GetBeginAtomIdx() == n else True
                    bond_feature_matrix[bond_index] = self.bond_tokenizer(
                        self.bond_features(bond, flipped=rev))
                    try:
                        bond_index_matrix[bond_index] = bond_index_array.tolist().index(bond.GetIdx())
                    except:
                        pass

                distance = distance_matrix[n, neighbor]
                bond_distance_matrix[bond_index] = distance
                 
                # update connectivity matrix
                connectivity[bond_index, 0] = n
                connectivity[bond_index, 1] = neighbor
                
                bond_index += 1
        return {
            'n_atom': n_atom,
            'n_bond': n_bond,
            'n_pro': n_pro,
            'atom': atom_feature_matrix,
            'bond': bond_feature_matrix,
            'distance': bond_distance_matrix,
            'connectivity': connectivity,
            'bond_index': bond_index_matrix,
        }

class MolAPreprocessor(MolPreprocessor):
    """
    This is a subclass of Molpreprocessor that preprocessor molecule with
    bond property target
    """
    def __init__(self, **kwargs):
        """
        A preprocessor class that also returns bond_target_matrix, besides the bond matrix
        returned by MolPreprocessor. The bond_target_matrix is then used as ref to reduce molecule
        to bond property
        """
        super(MolAPreprocessor, self).__init__(**kwargs)

    def construct_feature_matrices(self, entry):
        """
        Given an entry contining rdkit molecule, bond_index and for the target property, 
        return atom 
        feature matrices, bond feature matrices, distance matrices, connectivity matrices and bond
        ref matrices.

        returns
        dict with entries
        see MolPreproccessor
        'bond_index' : ref array to the bond index
        """
        mol, atom_index_array = entry
        
        n_atom = len(mol.GetAtoms())
        n_pro = len(atom_index_array)

        # n_bond is actually the number of atom-atom pairs, so this is defined
        # by the number of neighbors for each atom.
        #if there is cutoff, 
        distance_matrix = Chem.Get3DDistanceMatrix(mol)

        if self.n_neighbors <= (n_atom - 1):
            n_bond = self.n_neighbors * n_atom
        else:
            # If there are fewer atoms than n_neighbors, all atoms will be
            # connected
            n_bond = distance_matrix[(distance_matrix < self.cutoff) & (distance_matrix != 0)].size

        if n_bond == 0: n_bond = 1

        # Initialize the matrices to be filled in during the following loop.
        atom_feature_matrix = np.zeros(n_atom, dtype='int')
        bond_feature_matrix = np.zeros(n_bond, dtype='int')
        bond_distance_matrix = np.zeros(n_bond, dtype=np.float32)
        atom_index_matrix = np.full(n_atom, -1, dtype='int')
        connectivity = np.zeros((n_bond, 2), dtype='int')

        # Hopefully we've filtered out all problem mols by now.
        if mol is None:
            raise RuntimeError("Issue in loading mol")
        
        # Get a list of the atoms in the molecule.
        atom_seq = mol.GetAtoms()
        atoms = [atom_seq[i] for i in range(n_atom)]

        # Here we loop over each atom, and the inner loop iterates over each
        # neighbor of the current atom.
        bond_index = 0  # keep track of our current bond.
        for n, atom in enumerate(atoms):
            # update atom feature matrix
            atom_feature_matrix[n] = self.atom_tokenizer(
                self.atom_features(atom))
            try:
                atom_index_matrix[n] = atom_index_array.tolist().index(atom.GetIdx())
            except:
                pass 
            # if n_neighbors is greater than total atoms, then each atom is a
            # neighbor.
            if (self.n_neighbors + 1) > len(mol.GetAtoms()):
                neighbor_end_index = len(mol.GetAtoms())
            else:
                neighbor_end_index = (self.n_neighbors + 1)

            distance_atom = distance_matrix[n, :]
            cutoff_end_index = distance_atom[distance_atom < self.cutoff].size

            end_index = min(neighbor_end_index, cutoff_end_index)

            # Loop over each of the nearest neighbors

            neighbor_inds = distance_matrix[n, :].argsort()[1:end_index]
            if len(neighbor_inds)==0: neighbor_inds = [n]
            for neighbor in neighbor_inds:
                
                # update bond feature matrix
                bond = mol.GetBondBetweenAtoms(n, int(neighbor))
                if bond is None:
                    bond_feature_matrix[bond_index] = 0
                else:
                    rev = False if bond.GetBeginAtomIdx() == n else True
                    bond_feature_matrix[bond_index] = self.bond_tokenizer(
                        self.bond_features(bond, flipped=rev))

                distance = distance_matrix[n, neighbor]
                bond_distance_matrix[bond_index] = distance
                 
                # update connectivity matrix
                connectivity[bond_index, 0] = n
                connectivity[bond_index, 1] = neighbor
                
                bond_index += 1
        return {
            'n_atom': n_atom,
            'n_bond': n_bond,
            'n_pro': n_pro,
            'atom': atom_feature_matrix,
            'bond': bond_feature_matrix,
            'distance': bond_distance_matrix,
            'connectivity': connectivity,
            'atom_index': atom_index_matrix,
        }


# TODO: rewrite this                                
# class LaplacianSmilesPreprocessor(SmilesPreprocessor):
#     """ Extends the SmilesPreprocessor class to also return eigenvalues and
#     eigenvectors of the graph laplacian matrix.
#
#     Example:
#     >>> preprocessor = SmilesPreprocessor(
#     >>>     max_atoms=55, max_bonds=62, max_degree=4, explicit_hs=False)
#     >>> atom, connectivity, eigenvalues, eigenvectors = preprocessor.fit(
#             data.smiles)
#     """
#
#     def preprocess(self, smiles_iterator, train=True):
#
#         self.atom_tokenizer.train = train
#         self.bond_tokenizer.train = train
#
#         for smiles in tqdm(smiles_iterator):
#             G = self._mol_to_nx(smiles)
#             A = self._get_atom_feature_matrix(G)
#             C = self._get_connectivity_matrix(G)
#             W, V = self._get_laplacian_spectral_decomp(G)
#             yield A, C, W, V
#
#
#     def _get_laplacian_spectral_decomp(self, G):
#         """ Return the eigenvalues and eigenvectors of the graph G, padded to
#         `self.max_atoms`.
#         """
#
#         w0 = np.zeros((self.max_atoms, 1))
#         v0 = np.zeros((self.max_atoms, self.max_atoms))
#
#         w, v = eigh(nx.laplacian_matrix(G).todense())
#
#         num_atoms = len(v)
#
#         w0[:num_atoms, 0] = w
#         v0[:num_atoms, :num_atoms] = v
#
#         return w0, v0
#
#
#     def fit(self, smiles_iterator):
#         results = self._fit(smiles_iterator)
#         return {'atom': results[0], 
#                 'connectivity': results[1],
#                 'w': results[2],
#                 'v': results[3]}
#
#
#     def predict(self, smiles_iterator):
#         results = self._predict(smiles_iterator)
#         return {'atom': results[0], 
#                 'connectivity': results[1],
#                 'w': results[2],
#                 'v': results[3]}


def get_max_atom_bond_size(smiles_iterator, explicit_hs=True):
    """ Convienence function to get max_atoms, max_bonds for a set of input
    SMILES """

    max_atoms = 0
    max_bonds = 0
    for smiles in tqdm(smiles_iterator):
        mol = MolFromSmiles(smiles)
        if explicit_hs:
            mol = AddHs(mol)
        max_atoms = max([max_atoms, len(mol.GetAtoms())])
        max_bonds = max([max_bonds, len(mol.GetBonds())])

    return dict(max_atoms=max_atoms, max_bonds=max_bonds*2)


def canonicalize_smiles(smiles, isomeric=True, sanitize=True):
    try:
        mol = MolFromSmiles(smiles, sanitize=sanitize)
        return MolToSmiles(mol, isomericSmiles=isomeric)
    except Exception:
        pass