When SMILES Smiles, Practicality Judgment and Yield Prediction of Chemical Reaction via Deep Chemical Language Processing

Jiang, Shu; Zhang, Zhuosheng; Zhao, Hai; Li, Jiangtong; Yang, Yang; Lu, Bao-Liang; Ning, Xia

doi:10.1109/access.2021.3083838

Cited by 25 publications

(25 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Openly available data sets are derived from either patents [127,128] or chemical journals [129], and more rarely experimental procedures directly [130]. These data sets are distributed using SMILES as a representation for the reaction itself and usually include extra information in various formats.…”

Section: Reactionsmentioning

confidence: 99%

SELFIES and the future of molecular string representations

Krenn¹,

Ai²,

Barthel³

et al. 2022

Preprint

View full text Add to dashboard Cite

Artificial intelligence (AI) and machine learning (ML) are expanding in popularity for broad applications to challenging tasks in chemistry and materials science. Examples include the prediction of properties, the discovery of new reaction pathways, or the design of new molecules. The machine needs to read and write fluently in a chemical language for each of these tasks. Strings are a common tool to represent molecular graphs, and the most popular molecular string representation, SMILES, has powered cheminformatics since the late 1980s. However, in the context of AI and ML in chemistry, SMILES has several shortcomings -most pertinently, most combinations of symbols lead to invalid results with no valid chemical interpretation. To overcome this issue, a new language for molecules was introduced in 2020 that guarantees 100% robustness: SELFIES (SELF-referencIng Embedded Strings). SELFIES has since simplified and enabled numerous new applications in chemistry. In this manuscript, we look to the future and discuss molecular string representations, along with their respective opportunities and challenges. We propose 16 concrete Future Projects for

show abstract

Section: Reactionsmentioning

confidence: 99%

SELFIES and the future of molecular string representations

Krenn¹,

Ai²,

Barthel³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…With the advances of computing power, data availability, and algorithms, there has been significant interest in developing machine learning (ML) models to assist a variety of organic reaction-related tasks, 1−4 including reaction product prediction, 5−18 retrosynthesis, 9,14,17,19−37 reaction condition optimization, 38−41 reaction yield prediction, 42−54 and reaction type classification. 38,51,55−57 These ML-based data-driven approaches for organic synthesis can be classified into descriptor-based models, [5][6][7][8][9][10][19][20][21][22][23][24][25][26][38][39][40][41]51,55,57 graph-based m o d e l s , 1 1 − 1 3 , 2 7 , 2 8 , 5 2 a n d s e q u e n c e -b a s e d m o dels, [14][15][16][17][18][29][30][31][32][33][34][35][36][37]53,54,56 depending on how molecules are represented as input for machine learning. Descriptor-based models use hand-crafted features as molecular representations and often need feature engineering or template extraction for different reaction prediction tasks, which set limitations to generalizability.…”

Section: ■ Introductionmentioning

confidence: 99%

Unified Deep Learning Model for Multitask Reaction Predictions with Explanation

Zhang

2022

J. Chem. Inf. Model.

View full text Add to dashboard Cite

There is significant interest and importance to develop robust machine learning models to assist organic chemistry synthesis. Typically, task-specific machine learning models for distinct reaction prediction tasks have been developed. In this work, we develop a unified deep learning model, T5Chem, for a variety of chemical reaction predictions tasks by adapting the “Text-to-Text Transfer Transformer” (T5) framework in natural language processing (NLP). On the basis of self-supervised pretraining with PubChem molecules, the T5Chem model can achieve state-of-the-art performances for four distinct types of task-specific reaction prediction tasks using four different open-source data sets, including reaction type classification on USPTO_TPL, forward reaction prediction on USPTO_MIT, single-step retrosynthesis on USPTO_50k, and reaction yield prediction on high-throughput C–N coupling reactions. Meanwhile, we introduced a new unified multitask reaction prediction data set USPTO_500_MT, which can be used to train and test five different types of reaction tasks, including the above four as well as a new reagent suggestion task. Our results showed that models trained with multiple tasks are more robust and can benefit from mutual learning on related tasks. Furthermore, we demonstrated the use of SHAP (SHapley Additive exPlanations) to explain T5Chem predictions at the functional group level, which provides a way to demystify sequence-based deep learning models in chemistry. T5Chem is accessible through .

show abstract

“…In fact, interesting applications of NLP-based methods to chemical reactions are now becoming available. 29–31 The use of NLP-based models for accurate prediction of various properties of molecules is well-known. 32,33 On the other hand, predicting the reaction outcome that is known to depend on the molecular attributes of catalysts, reactants, solvents and several other factors is challenging and has seldom been reported using language models.…”

Section: Introductionmentioning

confidence: 99%