Discovery of novel chemical reactions by deep generative recurrent neural network
Abstract The “creativity” of Artificial Intelligence (AI) in terms of generating de novo molecular structures opened a novel paradigm in compound design, weaknesses (stability & feasibility issues of such structures) notwithstanding. Here we show that “creative” AI may be as successfully taught...
| Main Authors: | , , , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Nature Portfolio
2021-02-01
|
| Series: | Scientific Reports |
| Online Access: | https://doi.org/10.1038/s41598-021-81889-y |
| _version_ | 1818570750138777600 |
|---|---|
| author | William Bort Igor I. Baskin Timur Gimadiev Artem Mukanov Ramil Nugmanov Pavel Sidorov Gilles Marcou Dragos Horvath Olga Klimchuk Timur Madzhidov Alexandre Varnek |
| author_facet | William Bort Igor I. Baskin Timur Gimadiev Artem Mukanov Ramil Nugmanov Pavel Sidorov Gilles Marcou Dragos Horvath Olga Klimchuk Timur Madzhidov Alexandre Varnek |
| author_sort | William Bort |
| collection | DOAJ |
| description | Abstract The “creativity” of Artificial Intelligence (AI) in terms of generating de novo molecular structures opened a novel paradigm in compound design, weaknesses (stability & feasibility issues of such structures) notwithstanding. Here we show that “creative” AI may be as successfully taught to enumerate novel chemical reactions that are stoichiometrically coherent. Furthermore, when coupled to reaction space cartography, de novo reaction design may be focused on the desired reaction class. A sequence-to-sequence autoencoder with bidirectional Long Short-Term Memory layers was trained on on-purpose developed “SMILES/CGR” strings, encoding reactions of the USPTO database. The autoencoder latent space was visualized on a generative topographic map. Novel latent space points were sampled around a map area populated by Suzuki reactions and decoded to corresponding reactions. These can be critically analyzed by the expert, cleaned of irrelevant functional groups and eventually experimentally attempted, herewith enlarging the synthetic purpose of popular synthetic pathways. |
| first_indexed | 2024-12-14T13:46:01Z |
| format | Article |
| id | doaj.art-3d91d19a63b3499ebdae5082b570347c |
| institution | Directory Open Access Journal |
| issn | 2045-2322 |
| language | English |
| last_indexed | 2024-12-14T13:46:01Z |
| publishDate | 2021-02-01 |
| publisher | Nature Portfolio |
| record_format | Article |
| series | Scientific Reports |
| spelling | doaj.art-3d91d19a63b3499ebdae5082b570347c2022-12-21T22:59:18ZengNature PortfolioScientific Reports2045-23222021-02-0111111510.1038/s41598-021-81889-yDiscovery of novel chemical reactions by deep generative recurrent neural networkWilliam Bort0Igor I. Baskin1Timur Gimadiev2Artem Mukanov3Ramil Nugmanov4Pavel Sidorov5Gilles Marcou6Dragos Horvath7Olga Klimchuk8Timur Madzhidov9Alexandre Varnek10Laboratory of Chemoinformatics, UMR 7140 CNRS, University of StrasbourgLaboratory of Chemoinformatics, UMR 7140 CNRS, University of StrasbourgInstitute for Chemical Reaction Design and Discovery (WPI-ICReDD), Hokkaido UniversityLaboratory of Chemoinformatics and Molecular Modeling, Butlerov Institute of Chemistry, Kazan Federal UniversityLaboratory of Chemoinformatics and Molecular Modeling, Butlerov Institute of Chemistry, Kazan Federal UniversityInstitute for Chemical Reaction Design and Discovery (WPI-ICReDD), Hokkaido UniversityLaboratory of Chemoinformatics, UMR 7140 CNRS, University of StrasbourgLaboratory of Chemoinformatics, UMR 7140 CNRS, University of StrasbourgLaboratory of Chemoinformatics, UMR 7140 CNRS, University of StrasbourgLaboratory of Chemoinformatics and Molecular Modeling, Butlerov Institute of Chemistry, Kazan Federal UniversityLaboratory of Chemoinformatics, UMR 7140 CNRS, University of StrasbourgAbstract The “creativity” of Artificial Intelligence (AI) in terms of generating de novo molecular structures opened a novel paradigm in compound design, weaknesses (stability & feasibility issues of such structures) notwithstanding. Here we show that “creative” AI may be as successfully taught to enumerate novel chemical reactions that are stoichiometrically coherent. Furthermore, when coupled to reaction space cartography, de novo reaction design may be focused on the desired reaction class. A sequence-to-sequence autoencoder with bidirectional Long Short-Term Memory layers was trained on on-purpose developed “SMILES/CGR” strings, encoding reactions of the USPTO database. The autoencoder latent space was visualized on a generative topographic map. Novel latent space points were sampled around a map area populated by Suzuki reactions and decoded to corresponding reactions. These can be critically analyzed by the expert, cleaned of irrelevant functional groups and eventually experimentally attempted, herewith enlarging the synthetic purpose of popular synthetic pathways.https://doi.org/10.1038/s41598-021-81889-y |
| spellingShingle | William Bort Igor I. Baskin Timur Gimadiev Artem Mukanov Ramil Nugmanov Pavel Sidorov Gilles Marcou Dragos Horvath Olga Klimchuk Timur Madzhidov Alexandre Varnek Discovery of novel chemical reactions by deep generative recurrent neural network Scientific Reports |
| title | Discovery of novel chemical reactions by deep generative recurrent neural network |
| title_full | Discovery of novel chemical reactions by deep generative recurrent neural network |
| title_fullStr | Discovery of novel chemical reactions by deep generative recurrent neural network |
| title_full_unstemmed | Discovery of novel chemical reactions by deep generative recurrent neural network |
| title_short | Discovery of novel chemical reactions by deep generative recurrent neural network |
| title_sort | discovery of novel chemical reactions by deep generative recurrent neural network |
| url | https://doi.org/10.1038/s41598-021-81889-y |
| work_keys_str_mv | AT williambort discoveryofnovelchemicalreactionsbydeepgenerativerecurrentneuralnetwork AT igoribaskin discoveryofnovelchemicalreactionsbydeepgenerativerecurrentneuralnetwork AT timurgimadiev discoveryofnovelchemicalreactionsbydeepgenerativerecurrentneuralnetwork AT artemmukanov discoveryofnovelchemicalreactionsbydeepgenerativerecurrentneuralnetwork AT ramilnugmanov discoveryofnovelchemicalreactionsbydeepgenerativerecurrentneuralnetwork AT pavelsidorov discoveryofnovelchemicalreactionsbydeepgenerativerecurrentneuralnetwork AT gillesmarcou discoveryofnovelchemicalreactionsbydeepgenerativerecurrentneuralnetwork AT dragoshorvath discoveryofnovelchemicalreactionsbydeepgenerativerecurrentneuralnetwork AT olgaklimchuk discoveryofnovelchemicalreactionsbydeepgenerativerecurrentneuralnetwork AT timurmadzhidov discoveryofnovelchemicalreactionsbydeepgenerativerecurrentneuralnetwork AT alexandrevarnek discoveryofnovelchemicalreactionsbydeepgenerativerecurrentneuralnetwork |