selfies - npm Package Compare versions

Comparing version 1.0.3 to 1.0.4

PKG-INFO

 Metadata-Version: 2.1
 Name: selfies
-Version: 1.0.3
+Version: 1.0.4
 Summary: SELFIES (SELF-referencIng Embedded Strings) is a general-purpose, sequence-based, robust representation of semantically constrained graphs.
@@ -23,3 +23,5 @@ Home-page: https://github.com/aspuru-guzik-group/selfies
         [*Machine Learning: Science and Technology* **1**, 045024 (2020)](https://iopscience.iop.org/article/10.1088/2632-2153/aba947), [extensive blog post January 2021](https://aspuru.substack.com/p/molecular-graph-representations-and).<br>
-        Major contributors since v1.0.0: _[Alston Lo](https://github.com/aspuru-guzik-group/selfies/commits?author=alstonlo) and [Seyone Chithrananda](https://github.com/seyonechithrananda)_
+        [Talk on youtube about SELFIES](https://www.youtube.com/watch?v=CaIyUmfGXDk).<br>
+        Major contributors since v1.0.0: _[Alston Lo](https://github.com/aspuru-guzik-group/selfies/commits?author=alstonlo) and [Seyone Chithrananda](https://github.com/seyonechithrananda)_<br>
+        Chemistry Advisor: [Robert Pollice](https://scholar.google.at/citations?user=JR2N3JIAAAAJ)
         
@@ -144,8 +146,24 @@ A main objective is to use SELFIES as direct input into machine learning models,<br>
         * More examples can be found in the ``examples/`` directory, including a
-        variational autoencoder that runs on the SELFIES language.
+        [variational autoencoder that runs on the SELFIES](https://github.com/aspuru-guzik-group/selfies/tree/master/examples/vae_example) language.
         * This [ICLR2020 paper](https://arxiv.org/abs/1909.11655) used SELFIES in a
         genetic algorithm to achieve state-of-the-art performance for inverse design,
         with the [code here](https://github.com/aspuru-guzik-group/GA).
+        * SELFIES allows for [highly efficient exploration and interpolation of the chemical space](https://chemrxiv.org/articles/preprint/Beyond_Generative_Models_Superfast_Traversal_Optimization_Novelty_Exploration_and_Discovery_STONED_Algorithm_for_Molecules_using_SELFIES/13383266), with a [deterministic algorithms, see code](https://github.com/aspuru-guzik-group/stoned-selfies).
+        * We use SELFIES for [Deep Molecular dreaming](https://arxiv.org/abs/2012.09712), a new generative model inspired by interpretable neural networks in computational vision. See the [code of PASITHEA here](https://github.com/aspuru-guzik-group/Pasithea).
+        * Kohulan Rajan, Achim Zielesny, Christoph Steinbeck show in two papers that SELFIES outperforms other representations in [img2string](https://link.springer.com/article/10.1186/s13321-020-00469-w) and [string2string](https://chemrxiv.org/articles/preprint/STOUT_SMILES_to_IUPAC_Names_Using_Neural_Machine_Translation/13469202/1) translation tasks, see the codes of [DECIMER](https://github.com/Kohulan/DECIMER-Image-to-SMILES) and [STOUT](https://github.com/Kohulan/Smiles-TO-iUpac-Translator). 
         
         
+        ## Handling invalid inputs
+        If an invalid input is presented to the encoder or decoder, the return value is `None`.
+        The error can be analysed by using the `encoder(...,print_error=True)` option.
+        ```python
+        import selfies as sf
+        invalid_smiles="C[C@H](O)[C@@(*)C1=CC=CC=C1"
+        selfies_string=sf.encoder(invalid_smiles) 
+        
+        if selfies_string==None:
+            selfies_string=sf.encoder(invalid_smiles,print_error=True) 
+            # 'Encoding error 'C[C@H](O)[C@@(*)C1=CC=CC=C1': wildcard atom '*' not supported.'
+        ```
+        
         ## Tests
@@ -190,4 +208,5 @@ SELFIES uses `pytest` with `tox` as its testing framework.
         We thank Jacques Boitreaud, Andrew Brereton, Matthew Carbone (x94carbone), Nathan Frey (ncfrey), Theophile Gaudin,
-        Hyunmin Kim (hmkim), Minjie Li, Vincent Mallet, Kevin Ryan (LeanAndMean),  Benjamin Sanchez-Lengeling,
-        and Zhenpeng Yao for their suggestions and bug reports, and Robert Pollice for chemistry advices.
+        HelloJocelynLu, Hyunmin Kim (hmkim), Minjie Li, Vincent Mallet, Alexander Minidis (DocMinus), Kevin Ryan (LeanAndMean),
+        Benjamin Sanchez-Lengeling, and Zhenpeng Yao for their suggestions and bug reports,
+        and Robert Pollice for chemistry advices.
         
@@ -194,0 +213,0 @@ ## License

README.md

@@ -15,3 +15,5 @@ # SELFIES
 [*Machine Learning: Science and Technology* **1**, 045024 (2020)](https://iopscience.iop.org/article/10.1088/2632-2153/aba947), [extensive blog post January 2021](https://aspuru.substack.com/p/molecular-graph-representations-and).<br>
-Major contributors since v1.0.0: _[Alston Lo](https://github.com/aspuru-guzik-group/selfies/commits?author=alstonlo) and [Seyone Chithrananda](https://github.com/seyonechithrananda)_
+[Talk on youtube about SELFIES](https://www.youtube.com/watch?v=CaIyUmfGXDk).<br>
+Major contributors since v1.0.0: _[Alston Lo](https://github.com/aspuru-guzik-group/selfies/commits?author=alstonlo) and [Seyone Chithrananda](https://github.com/seyonechithrananda)_<br>
+Chemistry Advisor: [Robert Pollice](https://scholar.google.at/citations?user=JR2N3JIAAAAJ)
 
@@ -136,8 +138,24 @@ A main objective is to use SELFIES as direct input into machine learning models,<br>
 * More examples can be found in the ``examples/`` directory, including a
-variational autoencoder that runs on the SELFIES language.
+[variational autoencoder that runs on the SELFIES](https://github.com/aspuru-guzik-group/selfies/tree/master/examples/vae_example) language.
 * This [ICLR2020 paper](https://arxiv.org/abs/1909.11655) used SELFIES in a
 genetic algorithm to achieve state-of-the-art performance for inverse design,
 with the [code here](https://github.com/aspuru-guzik-group/GA).
+* SELFIES allows for [highly efficient exploration and interpolation of the chemical space](https://chemrxiv.org/articles/preprint/Beyond_Generative_Models_Superfast_Traversal_Optimization_Novelty_Exploration_and_Discovery_STONED_Algorithm_for_Molecules_using_SELFIES/13383266), with a [deterministic algorithms, see code](https://github.com/aspuru-guzik-group/stoned-selfies).
+* We use SELFIES for [Deep Molecular dreaming](https://arxiv.org/abs/2012.09712), a new generative model inspired by interpretable neural networks in computational vision. See the [code of PASITHEA here](https://github.com/aspuru-guzik-group/Pasithea).
+* Kohulan Rajan, Achim Zielesny, Christoph Steinbeck show in two papers that SELFIES outperforms other representations in [img2string](https://link.springer.com/article/10.1186/s13321-020-00469-w) and [string2string](https://chemrxiv.org/articles/preprint/STOUT_SMILES_to_IUPAC_Names_Using_Neural_Machine_Translation/13469202/1) translation tasks, see the codes of [DECIMER](https://github.com/Kohulan/DECIMER-Image-to-SMILES) and [STOUT](https://github.com/Kohulan/Smiles-TO-iUpac-Translator). 
 
 
+## Handling invalid inputs
+If an invalid input is presented to the encoder or decoder, the return value is `None`.
+The error can be analysed by using the `encoder(...,print_error=True)` option.
+```python
+import selfies as sf
+invalid_smiles="C[C@H](O)[C@@(*)C1=CC=CC=C1"
+selfies_string=sf.encoder(invalid_smiles) 
+
+if selfies_string==None:
+    selfies_string=sf.encoder(invalid_smiles,print_error=True) 
+    # 'Encoding error 'C[C@H](O)[C@@(*)C1=CC=CC=C1': wildcard atom '*' not supported.'
+```
+
 ## Tests
@@ -182,4 +200,5 @@ SELFIES uses `pytest` with `tox` as its testing framework.
 We thank Jacques Boitreaud, Andrew Brereton, Matthew Carbone (x94carbone), Nathan Frey (ncfrey), Theophile Gaudin,
-Hyunmin Kim (hmkim), Minjie Li, Vincent Mallet, Kevin Ryan (LeanAndMean),  Benjamin Sanchez-Lengeling,
-and Zhenpeng Yao for their suggestions and bug reports, and Robert Pollice for chemistry advices.
+HelloJocelynLu, Hyunmin Kim (hmkim), Minjie Li, Vincent Mallet, Alexander Minidis (DocMinus), Kevin Ryan (LeanAndMean),
+Benjamin Sanchez-Lengeling, and Zhenpeng Yao for their suggestions and bug reports,
+and Robert Pollice for chemistry advices.
 
@@ -186,0 +205,0 @@ ## License

selfies.egg-info/PKG-INFO

 Metadata-Version: 2.1
 Name: selfies
-Version: 1.0.3
+Version: 1.0.4
 Summary: SELFIES (SELF-referencIng Embedded Strings) is a general-purpose, sequence-based, robust representation of semantically constrained graphs.
@@ -23,3 +23,5 @@ Home-page: https://github.com/aspuru-guzik-group/selfies
         [*Machine Learning: Science and Technology* **1**, 045024 (2020)](https://iopscience.iop.org/article/10.1088/2632-2153/aba947), [extensive blog post January 2021](https://aspuru.substack.com/p/molecular-graph-representations-and).<br>
-        Major contributors since v1.0.0: _[Alston Lo](https://github.com/aspuru-guzik-group/selfies/commits?author=alstonlo) and [Seyone Chithrananda](https://github.com/seyonechithrananda)_
+        [Talk on youtube about SELFIES](https://www.youtube.com/watch?v=CaIyUmfGXDk).<br>
+        Major contributors since v1.0.0: _[Alston Lo](https://github.com/aspuru-guzik-group/selfies/commits?author=alstonlo) and [Seyone Chithrananda](https://github.com/seyonechithrananda)_<br>
+        Chemistry Advisor: [Robert Pollice](https://scholar.google.at/citations?user=JR2N3JIAAAAJ)
         
@@ -144,8 +146,24 @@ A main objective is to use SELFIES as direct input into machine learning models,<br>
         * More examples can be found in the ``examples/`` directory, including a
-        variational autoencoder that runs on the SELFIES language.
+        [variational autoencoder that runs on the SELFIES](https://github.com/aspuru-guzik-group/selfies/tree/master/examples/vae_example) language.
         * This [ICLR2020 paper](https://arxiv.org/abs/1909.11655) used SELFIES in a
         genetic algorithm to achieve state-of-the-art performance for inverse design,
         with the [code here](https://github.com/aspuru-guzik-group/GA).
+        * SELFIES allows for [highly efficient exploration and interpolation of the chemical space](https://chemrxiv.org/articles/preprint/Beyond_Generative_Models_Superfast_Traversal_Optimization_Novelty_Exploration_and_Discovery_STONED_Algorithm_for_Molecules_using_SELFIES/13383266), with a [deterministic algorithms, see code](https://github.com/aspuru-guzik-group/stoned-selfies).
+        * We use SELFIES for [Deep Molecular dreaming](https://arxiv.org/abs/2012.09712), a new generative model inspired by interpretable neural networks in computational vision. See the [code of PASITHEA here](https://github.com/aspuru-guzik-group/Pasithea).
+        * Kohulan Rajan, Achim Zielesny, Christoph Steinbeck show in two papers that SELFIES outperforms other representations in [img2string](https://link.springer.com/article/10.1186/s13321-020-00469-w) and [string2string](https://chemrxiv.org/articles/preprint/STOUT_SMILES_to_IUPAC_Names_Using_Neural_Machine_Translation/13469202/1) translation tasks, see the codes of [DECIMER](https://github.com/Kohulan/DECIMER-Image-to-SMILES) and [STOUT](https://github.com/Kohulan/Smiles-TO-iUpac-Translator). 
         
         
+        ## Handling invalid inputs
+        If an invalid input is presented to the encoder or decoder, the return value is `None`.
+        The error can be analysed by using the `encoder(...,print_error=True)` option.
+        ```python
+        import selfies as sf
+        invalid_smiles="C[C@H](O)[C@@(*)C1=CC=CC=C1"
+        selfies_string=sf.encoder(invalid_smiles) 
+        
+        if selfies_string==None:
+            selfies_string=sf.encoder(invalid_smiles,print_error=True) 
+            # 'Encoding error 'C[C@H](O)[C@@(*)C1=CC=CC=C1': wildcard atom '*' not supported.'
+        ```
+        
         ## Tests
@@ -190,4 +208,5 @@ SELFIES uses `pytest` with `tox` as its testing framework.
         We thank Jacques Boitreaud, Andrew Brereton, Matthew Carbone (x94carbone), Nathan Frey (ncfrey), Theophile Gaudin,
-        Hyunmin Kim (hmkim), Minjie Li, Vincent Mallet, Kevin Ryan (LeanAndMean),  Benjamin Sanchez-Lengeling,
-        and Zhenpeng Yao for their suggestions and bug reports, and Robert Pollice for chemistry advices.
+        HelloJocelynLu, Hyunmin Kim (hmkim), Minjie Li, Vincent Mallet, Alexander Minidis (DocMinus), Kevin Ryan (LeanAndMean),
+        Benjamin Sanchez-Lengeling, and Zhenpeng Yao for their suggestions and bug reports,
+        and Robert Pollice for chemistry advices.
         
@@ -194,0 +213,0 @@ ## License

selfies/__init__.py

@@ -34,2 +34,5 @@ #!/usr/bin/env python
     "get_semantic_robust_alphabet",
+    "get_default_constraints",
+    "get_octet_rule_constraints",
+    "get_hypervalent_constraints",
     "get_semantic_constraints",
@@ -50,2 +53,5 @@ "set_semantic_constraints",
     get_semantic_robust_alphabet,
+    get_default_constraints,
+    get_octet_rule_constraints,
+    get_hypervalent_constraints,
     get_semantic_constraints,
@@ -52,0 +58,0 @@ set_semantic_constraints,

selfies/decoder.py

 from collections import OrderedDict
 from typing import Dict, Iterable, List, Optional, Tuple, Union
 
-from selfies.grammar_rules import get_bond_from_num, get_n_from_symbols, \
-    get_next_branch_state, get_next_state, get_num_from_bond
+from selfies.grammar_rules import (get_bond_from_num,
+                                   get_hypervalent_constraints,
+                                   get_n_from_symbols, get_next_branch_state,
+                                   get_next_state, get_num_from_bond,
+                                   get_octet_rule_constraints,
+                                   get_semantic_constraints,
+                                   set_semantic_constraints)
 
 
-def decoder(selfies: str, print_error: bool = False) -> Optional[str]:
+def decoder(selfies: str,
+            print_error: bool = False,
+            constraints: Optional[str] = None) -> Optional[str]:
     """Translates a SELFIES into a SMILES.
@@ -22,2 +29,6 @@
         Defaults to False.
+    :param constraints: if ``'octet_rule'`` or ``'hypervalent'``,
+        the corresponding preset bond constraints will be used instead.
+        If ``None``, :func:`selfies.decoder` will use the
+        currently configured bond constraints. Defaults to ``None``.
     :return: the SMILES translation of ``selfies``. If an error occurs,
@@ -31,4 +42,29 @@ and ``selfies`` cannot be translated, ``None`` is returned instead.
     'C=CF'
+
+    .. seealso:: The
+        `"octet_rule" <https://en.wikipedia.org/wiki/Octet_rule>`_
+        and
+        `"hypervalent" <https://en.wikipedia.org/wiki/Hypervalent_molecule>`_
+        preset bond constraints
+        can be viewed with :func:`selfies.get_octet_rule_constraints` and
+        :func:`selfies.get_hypervalent_constraints`, respectively. These
+        presets are variants of the "default" bond constraints, which can
+        be viewed with :func:`selfies.get_default_constraints`. Their
+        differences can be summarized as follows:
+
+            * def. : ``Cl``, ``Br``, ``I``: 1, ``N``: 3, ``P``: 5, ``P+1``: 6, ``P-1``: 4, ``S``: 6, ``S+1``: 7, ``S-1``: 5
+            * oct. : ``Cl``, ``Br``, ``I``: 1, ``N``: 3, ``P``: 3, ``P+1``: 4, ``P-1``: 2, ``S``: 2, ``S+1``: 3, ``S-1``: 1
+            * hyp. : ``Cl``, ``Br``, ``I``: 7, ``N``: 5, ``P``: 5, ``P+1``: 6, ``P-1``: 4, ``S``: 6, ``S+1``: 7, ``S-1``: 5
     """
 
+    old_constraints = get_semantic_constraints()
+    if constraints is None:
+        pass
+    elif constraints == 'octet_rule':
+        set_semantic_constraints(get_octet_rule_constraints())
+    elif constraints == 'hypervalent':
+        set_semantic_constraints(get_hypervalent_constraints())
+    else:
+        raise ValueError("unrecognized constraint type")
+
     try:
@@ -43,5 +79,11 @@ all_smiles = []  # process dot-separated fragments separately
 
+        if constraints is not None:  # restore old constraints
+            set_semantic_constraints(old_constraints)
+
         return '.'.join(all_smiles)
 
     except ValueError as err:
+        if constraints is not None:  # restore old constraints
+            set_semantic_constraints(old_constraints)
+
         if print_error:
@@ -48,0 +90,0 @@ print("Decoding error '{}': {}.".format(selfies, err))

selfies/grammar_rules.py

@@ -9,7 +9,17 @@ from itertools import product
     'C': 4, 'C+1': 5, 'C-1': 3,
+    'P': 5, 'P+1': 6, 'P-1': 4,
     'S': 6, 'S+1': 7, 'S-1': 5,
-    'P': 7, 'P+1': 8, 'P-1': 6,
-    '?': 8,
+    '?': 8
 }
 
+octet_rule_bond_constraints = dict(default_bond_constraints)
+octet_rule_bond_constraints.update(
+    {'S': 2, 'S+1': 3, 'S-1': 1, 'P': 3, 'P+1': 4, 'P-1': 2}
+)
+
+hypervalent_bond_constraints = dict(default_bond_constraints)
+hypervalent_bond_constraints.update(
+    {'Cl': 7, 'Br': 7, 'I': 7, 'N': 5}
+)
+
 _bond_constraints = default_bond_constraints
@@ -57,2 +67,43 @@
 
+def get_default_constraints() -> Dict[str, int]:
+    """Returns the preset "default" bond constraint settings.
+
+    :return: the default constraint settings.
+    """
+
+    global default_bond_constraints
+    return dict(default_bond_constraints)
+
+
+def get_octet_rule_constraints() -> Dict[str, int]:
+    """Returns the preset "octet rule" bond constraint settings. These
+    constraints are a harsher version of the default constraints, so that
+    the `octet rule <https://en.wikipedia.org/wiki/Octet_rule>`_
+    is obeyed. In particular, ``S`` and ``P`` are
+    restricted to a 2 and 3 bond capacity, respectively (and similarly with
+    ``S+``, ``S-``, ``P+``, ``P-``).
+
+    :return: the octet rule constraint settings.
+    """
+
+    global octet_rule_bond_constraints
+    return dict(octet_rule_bond_constraints)
+
+
+def get_hypervalent_constraints() -> Dict[str, int]:
+    """Returns the preset "hypervalent" bond constraint settings. These
+    constraints are a relaxed version of the default constraints, to allow
+    for `hypervalent molecules
+    <https://en.wikipedia.org/wiki/Hypervalent_molecule>`_.
+    In particular, ``Cl``, ``Br``, and ``I``
+    are relaxed to a 7 bond capacity, and ``N`` is relaxed to a 5 bond
+    capacity.
+
+    :return: the hypervalent constraint settings.
+    """
+
+    global hypervalent_bond_constraints
+    return dict(hypervalent_bond_constraints)
+
+
 def get_semantic_constraints() -> Dict[str, int]:
@@ -59,0 +110,0 @@ """Returns the semantic bond constraints that :mod:`selfies` is currently

setup.py

@@ -10,3 +10,3 @@ #!/usr/bin/env python
     name="selfies",
-    version="1.0.3",
+    version="1.0.4",
     author="Mario Krenn",
@@ -13,0 +13,0 @@ author_email="[e-mail], [e-mail]",

Alert delta unavailable

Currently unable to show alert delta for PyPI packages.

Improved metrics

Total package byte prevSize

106161

10.72%

Lines of code

1475

5.73%