Using Machine Learning To Predict Suitable Conditions for Organic Reactions
© Copyright 2018 American Chemical Society. Reaction condition recommendation is an essential element for the realization of computer-assisted synthetic planning. Accurate suggestions of reaction conditions are required for experimental validation and can have a significant effect on the success or...
| Main Authors: | , , , , , |
|---|---|
| Other Authors: | |
| Format: | Article |
| Language: | English |
| Published: |
American Chemical Society (ACS)
2021
|
| Online Access: | https://hdl.handle.net/1721.1/135864 |
| _version_ | 1826205116622962688 |
|---|---|
| author | Gao, Hanyu Struble, Thomas J Coley, Connor W Wang, Yuran Green, William H Jensen, Klavs F |
| author2 | Massachusetts Institute of Technology. Department of Chemical Engineering |
| author_facet | Massachusetts Institute of Technology. Department of Chemical Engineering Gao, Hanyu Struble, Thomas J Coley, Connor W Wang, Yuran Green, William H Jensen, Klavs F |
| author_sort | Gao, Hanyu |
| collection | MIT |
| description | © Copyright 2018 American Chemical Society. Reaction condition recommendation is an essential element for the realization of computer-assisted synthetic planning. Accurate suggestions of reaction conditions are required for experimental validation and can have a significant effect on the success or failure of an attempted transformation. However, de novo condition recommendation remains a challenging and under-explored problem and relies heavily on chemists' knowledge and experience. In this work, we develop a neural-network model to predict the chemical context (catalyst(s), solvent(s), reagent(s)), as well as the temperature most suitable for any particular organic reaction. Trained on ∼10 million examples from Reaxys, the model is able to propose conditions where a close match to the recorded catalyst, solvent, and reagent is found within the top-10 predictions 69.6% of the time, with top-10 accuracies for individual species reaching 80-90%. Temperature is accurately predicted within ±20 °C from the recorded temperature in 60-70% of test cases, with higher accuracy for cases with correct chemical context predictions. The utility of the model is illustrated through several examples spanning a range of common reaction classes. We also demonstrate that the model implicitly learns a continuous numerical embedding of solvent and reagent species that captures their functional similarity. |
| first_indexed | 2024-09-23T13:07:19Z |
| format | Article |
| id | mit-1721.1/135864 |
| institution | Massachusetts Institute of Technology |
| language | English |
| last_indexed | 2024-09-23T13:07:19Z |
| publishDate | 2021 |
| publisher | American Chemical Society (ACS) |
| record_format | dspace |
| spelling | mit-1721.1/1358642023-03-15T20:17:21Z Using Machine Learning To Predict Suitable Conditions for Organic Reactions Gao, Hanyu Struble, Thomas J Coley, Connor W Wang, Yuran Green, William H Jensen, Klavs F Massachusetts Institute of Technology. Department of Chemical Engineering © Copyright 2018 American Chemical Society. Reaction condition recommendation is an essential element for the realization of computer-assisted synthetic planning. Accurate suggestions of reaction conditions are required for experimental validation and can have a significant effect on the success or failure of an attempted transformation. However, de novo condition recommendation remains a challenging and under-explored problem and relies heavily on chemists' knowledge and experience. In this work, we develop a neural-network model to predict the chemical context (catalyst(s), solvent(s), reagent(s)), as well as the temperature most suitable for any particular organic reaction. Trained on ∼10 million examples from Reaxys, the model is able to propose conditions where a close match to the recorded catalyst, solvent, and reagent is found within the top-10 predictions 69.6% of the time, with top-10 accuracies for individual species reaching 80-90%. Temperature is accurately predicted within ±20 °C from the recorded temperature in 60-70% of test cases, with higher accuracy for cases with correct chemical context predictions. The utility of the model is illustrated through several examples spanning a range of common reaction classes. We also demonstrate that the model implicitly learns a continuous numerical embedding of solvent and reagent species that captures their functional similarity. 2021-10-27T20:29:42Z 2021-10-27T20:29:42Z 2018 2019-08-19T17:44:53Z Article http://purl.org/eprint/type/JournalArticle https://hdl.handle.net/1721.1/135864 en 10.1021/ACSCENTSCI.8B00357 ACS Central Science Article is made available in accordance with the publisher's policy and may be subject to US copyright law. Please refer to the publisher's site for terms of use. application/pdf American Chemical Society (ACS) ACS |
| spellingShingle | Gao, Hanyu Struble, Thomas J Coley, Connor W Wang, Yuran Green, William H Jensen, Klavs F Using Machine Learning To Predict Suitable Conditions for Organic Reactions |
| title | Using Machine Learning To Predict Suitable Conditions for Organic Reactions |
| title_full | Using Machine Learning To Predict Suitable Conditions for Organic Reactions |
| title_fullStr | Using Machine Learning To Predict Suitable Conditions for Organic Reactions |
| title_full_unstemmed | Using Machine Learning To Predict Suitable Conditions for Organic Reactions |
| title_short | Using Machine Learning To Predict Suitable Conditions for Organic Reactions |
| title_sort | using machine learning to predict suitable conditions for organic reactions |
| url | https://hdl.handle.net/1721.1/135864 |
| work_keys_str_mv | AT gaohanyu usingmachinelearningtopredictsuitableconditionsfororganicreactions AT strublethomasj usingmachinelearningtopredictsuitableconditionsfororganicreactions AT coleyconnorw usingmachinelearningtopredictsuitableconditionsfororganicreactions AT wangyuran usingmachinelearningtopredictsuitableconditionsfororganicreactions AT greenwilliamh usingmachinelearningtopredictsuitableconditionsfororganicreactions AT jensenklavsf usingmachinelearningtopredictsuitableconditionsfororganicreactions |