SMILES-based deep generative scaffold decorator for de-novo drug design
Abstract Molecular generative models trained with small sets of molecules represented as SMILES strings can generate large regions of the chemical space. Unfortunately, due to the sequential nature of SMILES strings, these models are not able to generate molecules given a scaffold (i.e., partially-b...
| Main Authors: | , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
BMC
2020-05-01
|
| Series: | Journal of Cheminformatics |
| Subjects: | |
| Online Access: | http://link.springer.com/article/10.1186/s13321-020-00441-8 |
| _version_ | 1819111975153565696 |
|---|---|
| author | Josep Arús-Pous Atanas Patronov Esben Jannik Bjerrum Christian Tyrchan Jean-Louis Reymond Hongming Chen Ola Engkvist |
| author_facet | Josep Arús-Pous Atanas Patronov Esben Jannik Bjerrum Christian Tyrchan Jean-Louis Reymond Hongming Chen Ola Engkvist |
| author_sort | Josep Arús-Pous |
| collection | DOAJ |
| description | Abstract Molecular generative models trained with small sets of molecules represented as SMILES strings can generate large regions of the chemical space. Unfortunately, due to the sequential nature of SMILES strings, these models are not able to generate molecules given a scaffold (i.e., partially-built molecules with explicit attachment points). Herein we report a new SMILES-based molecular generative architecture that generates molecules from scaffolds and can be trained from any arbitrary molecular set. This approach is possible thanks to a new molecular set pre-processing algorithm that exhaustively slices all possible combinations of acyclic bonds of every molecule, combinatorically obtaining a large number of scaffolds with their respective decorations. Moreover, it serves as a data augmentation technique and can be readily coupled with randomized SMILES to obtain even better results with small sets. Two examples showcasing the potential of the architecture in medicinal and synthetic chemistry are described: First, models were trained with a training set obtained from a small set of Dopamine Receptor D2 (DRD2) active modulators and were able to meaningfully decorate a wide range of scaffolds and obtain molecular series predicted active on DRD2. Second, a larger set of drug-like molecules from ChEMBL was selectively sliced using synthetic chemistry constraints (RECAP rules). In this case, the resulting scaffolds with decorations were filtered only to allow those that included fragment-like decorations. This filtering process allowed models trained with this dataset to selectively decorate diverse scaffolds with fragments that were generally predicted to be synthesizable and attachable to the scaffold using known synthetic approaches. In both cases, the models were already able to decorate molecules using specific knowledge without the need to add it with other techniques, such as reinforcement learning. We envision that this architecture will become a useful addition to the already existent architectures for de novo molecular generation. |
| first_indexed | 2024-12-22T04:06:09Z |
| format | Article |
| id | doaj.art-c20116d0411b447eaa88d36151b88006 |
| institution | Directory Open Access Journal |
| issn | 1758-2946 |
| language | English |
| last_indexed | 2024-12-22T04:06:09Z |
| publishDate | 2020-05-01 |
| publisher | BMC |
| record_format | Article |
| series | Journal of Cheminformatics |
| spelling | doaj.art-c20116d0411b447eaa88d36151b880062022-12-21T18:39:37ZengBMCJournal of Cheminformatics1758-29462020-05-0112111810.1186/s13321-020-00441-8SMILES-based deep generative scaffold decorator for de-novo drug designJosep Arús-Pous0Atanas Patronov1Esben Jannik Bjerrum2Christian Tyrchan3Jean-Louis Reymond4Hongming Chen5Ola Engkvist6Molecular AI, Hit Discovery, Discovery Sciences, BioPharmaceutical R&D, AstraZenecaMolecular AI, Hit Discovery, Discovery Sciences, BioPharmaceutical R&D, AstraZenecaMolecular AI, Hit Discovery, Discovery Sciences, BioPharmaceutical R&D, AstraZenecaMedicinal Chemistry, Respiratory Inflammation, and Autoimmune (RIA), BioPharmaceutical R&D, AstraZenecaDepartment of Chemistry and Biochemistry, University of BernChemistry and Chemical Biology Centre, Guangzhou Regenerative Medicine and Health -Guangdong LaboratoryMolecular AI, Hit Discovery, Discovery Sciences, BioPharmaceutical R&D, AstraZenecaAbstract Molecular generative models trained with small sets of molecules represented as SMILES strings can generate large regions of the chemical space. Unfortunately, due to the sequential nature of SMILES strings, these models are not able to generate molecules given a scaffold (i.e., partially-built molecules with explicit attachment points). Herein we report a new SMILES-based molecular generative architecture that generates molecules from scaffolds and can be trained from any arbitrary molecular set. This approach is possible thanks to a new molecular set pre-processing algorithm that exhaustively slices all possible combinations of acyclic bonds of every molecule, combinatorically obtaining a large number of scaffolds with their respective decorations. Moreover, it serves as a data augmentation technique and can be readily coupled with randomized SMILES to obtain even better results with small sets. Two examples showcasing the potential of the architecture in medicinal and synthetic chemistry are described: First, models were trained with a training set obtained from a small set of Dopamine Receptor D2 (DRD2) active modulators and were able to meaningfully decorate a wide range of scaffolds and obtain molecular series predicted active on DRD2. Second, a larger set of drug-like molecules from ChEMBL was selectively sliced using synthetic chemistry constraints (RECAP rules). In this case, the resulting scaffolds with decorations were filtered only to allow those that included fragment-like decorations. This filtering process allowed models trained with this dataset to selectively decorate diverse scaffolds with fragments that were generally predicted to be synthesizable and attachable to the scaffold using known synthetic approaches. In both cases, the models were already able to decorate molecules using specific knowledge without the need to add it with other techniques, such as reinforcement learning. We envision that this architecture will become a useful addition to the already existent architectures for de novo molecular generation.http://link.springer.com/article/10.1186/s13321-020-00441-8Deep learningGenerative modelsSMILESRandomized SMILESRecurrent neural networksFragment-based drug discovery |
| spellingShingle | Josep Arús-Pous Atanas Patronov Esben Jannik Bjerrum Christian Tyrchan Jean-Louis Reymond Hongming Chen Ola Engkvist SMILES-based deep generative scaffold decorator for de-novo drug design Journal of Cheminformatics Deep learning Generative models SMILES Randomized SMILES Recurrent neural networks Fragment-based drug discovery |
| title | SMILES-based deep generative scaffold decorator for de-novo drug design |
| title_full | SMILES-based deep generative scaffold decorator for de-novo drug design |
| title_fullStr | SMILES-based deep generative scaffold decorator for de-novo drug design |
| title_full_unstemmed | SMILES-based deep generative scaffold decorator for de-novo drug design |
| title_short | SMILES-based deep generative scaffold decorator for de-novo drug design |
| title_sort | smiles based deep generative scaffold decorator for de novo drug design |
| topic | Deep learning Generative models SMILES Randomized SMILES Recurrent neural networks Fragment-based drug discovery |
| url | http://link.springer.com/article/10.1186/s13321-020-00441-8 |
| work_keys_str_mv | AT joseparuspous smilesbaseddeepgenerativescaffolddecoratorfordenovodrugdesign AT atanaspatronov smilesbaseddeepgenerativescaffolddecoratorfordenovodrugdesign AT esbenjannikbjerrum smilesbaseddeepgenerativescaffolddecoratorfordenovodrugdesign AT christiantyrchan smilesbaseddeepgenerativescaffolddecoratorfordenovodrugdesign AT jeanlouisreymond smilesbaseddeepgenerativescaffolddecoratorfordenovodrugdesign AT hongmingchen smilesbaseddeepgenerativescaffolddecoratorfordenovodrugdesign AT olaengkvist smilesbaseddeepgenerativescaffolddecoratorfordenovodrugdesign |