Improving the performance of models for one-step retrosynthesis through re-ranking
Abstract Retrosynthesis is at the core of organic chemistry. Recently, the rapid growth of artificial intelligence (AI) has spurred a variety of novel machine learning approaches for data-driven synthesis planning. These methods learn complex patterns from reaction databases in orde...
| Main Authors: | , , |
|---|---|
| Other Authors: | |
| Format: | Article |
| Language: | English |
| Published: |
Springer International Publishing
2022
|
| Online Access: | https://hdl.handle.net/1721.1/141316 |
| _version_ | 1826210958034337792 |
|---|---|
| author | Lin, Min H. Tu, Zhengkai Coley, Connor W. |
| author2 | Massachusetts Institute of Technology. Department of Chemical Engineering |
| author_facet | Massachusetts Institute of Technology. Department of Chemical Engineering Lin, Min H. Tu, Zhengkai Coley, Connor W. |
| author_sort | Lin, Min H. |
| collection | MIT |
| description | Abstract
Retrosynthesis is at the core of organic chemistry. Recently, the rapid growth of artificial intelligence (AI) has spurred a variety of novel machine learning approaches for data-driven synthesis planning. These methods learn complex patterns from reaction databases in order to predict, for a given product, sets of reactants that can be used to synthesise that product. However, their performance as measured by the top-N accuracy in matching published reaction precedents still leaves room for improvement. This work aims to enhance these models by learning to re-rank their reactant predictions. Specifically, we design and train an energy-based model to re-rank, for each product, the published reaction as the top suggestion and the remaining reactant predictions as lower-ranked. We show that re-ranking can improve one-step models significantly using the standard USPTO-50k benchmark dataset, such as RetroSim, a similarity-based method, from 35.7 to 51.8% top-1 accuracy and NeuralSym, a deep learning method, from 45.7 to 51.3%, and also that re-ranking the union of two models’ suggestions can lead to better performance than either alone. However, the state-of-the-art top-1 accuracy is not improved by this method.
Graphical Abstract |
| first_indexed | 2024-09-23T14:58:09Z |
| format | Article |
| id | mit-1721.1/141316 |
| institution | Massachusetts Institute of Technology |
| language | English |
| last_indexed | 2024-09-23T14:58:09Z |
| publishDate | 2022 |
| publisher | Springer International Publishing |
| record_format | dspace |
| spelling | mit-1721.1/1413162023-02-09T20:12:40Z Improving the performance of models for one-step retrosynthesis through re-ranking Lin, Min H. Tu, Zhengkai Coley, Connor W. Massachusetts Institute of Technology. Department of Chemical Engineering Abstract Retrosynthesis is at the core of organic chemistry. Recently, the rapid growth of artificial intelligence (AI) has spurred a variety of novel machine learning approaches for data-driven synthesis planning. These methods learn complex patterns from reaction databases in order to predict, for a given product, sets of reactants that can be used to synthesise that product. However, their performance as measured by the top-N accuracy in matching published reaction precedents still leaves room for improvement. This work aims to enhance these models by learning to re-rank their reactant predictions. Specifically, we design and train an energy-based model to re-rank, for each product, the published reaction as the top suggestion and the remaining reactant predictions as lower-ranked. We show that re-ranking can improve one-step models significantly using the standard USPTO-50k benchmark dataset, such as RetroSim, a similarity-based method, from 35.7 to 51.8% top-1 accuracy and NeuralSym, a deep learning method, from 45.7 to 51.3%, and also that re-ranking the union of two models’ suggestions can lead to better performance than either alone. However, the state-of-the-art top-1 accuracy is not improved by this method. Graphical Abstract 2022-03-21T12:56:08Z 2022-03-21T12:56:08Z 2022-03-15 2022-03-20T04:15:26Z Article http://purl.org/eprint/type/JournalArticle https://hdl.handle.net/1721.1/141316 Journal of Cheminformatics. 2022 Mar 15;14(1):15 PUBLISHER_CC en https://doi.org/10.1186/s13321-022-00594-8 Creative Commons Attribution https://creativecommons.org/licenses/by/4.0/ The Author(s) application/pdf Springer International Publishing Springer International Publishing |
| spellingShingle | Lin, Min H. Tu, Zhengkai Coley, Connor W. Improving the performance of models for one-step retrosynthesis through re-ranking |
| title | Improving the performance of models for one-step retrosynthesis through re-ranking |
| title_full | Improving the performance of models for one-step retrosynthesis through re-ranking |
| title_fullStr | Improving the performance of models for one-step retrosynthesis through re-ranking |
| title_full_unstemmed | Improving the performance of models for one-step retrosynthesis through re-ranking |
| title_short | Improving the performance of models for one-step retrosynthesis through re-ranking |
| title_sort | improving the performance of models for one step retrosynthesis through re ranking |
| url | https://hdl.handle.net/1721.1/141316 |
| work_keys_str_mv | AT linminh improvingtheperformanceofmodelsforonestepretrosynthesisthroughreranking AT tuzhengkai improvingtheperformanceofmodelsforonestepretrosynthesisthroughreranking AT coleyconnorw improvingtheperformanceofmodelsforonestepretrosynthesisthroughreranking |