Discovery of novel chemical reactions by deep generative recurrent neural network

Implementation of a soft grading system for chemistry in a Moodle plugin: reaction handling

Here, we present a new method for evaluating questions on chemical reactions in the context of remote education. This method can be used when binary grading is not sufficient as some tolerance may be acceptable. In order to determine a grade, the developed workflow uses the pairwise similarity assessment of two considered reactions, each encoded by a single molecular graph with the help of the Condensed Graph of Reaction (CGR) approach. This workflow is part of the ChemMoodle project and is implemented as a Moodle Plugin. It uses the Chemdoodle engine for reaction drawing and visualization and communicates with a REST server calculating the similarity score using ISIDA fragment descriptors. The plugin is open-source, accessible in GitHub ( https://github.com/Laboratoire-de-Chemoinformatique/moodle-qtype_reacsimilarity ) and on the Moodle plugin store ( https://moodle.org/plugins/qtype_reacsimilarity?lang=en ). Both similarity measures and fragmentation can be configured.Scientific contribution This work introduces an open-source method for evaluating chemical reaction questions within Moodle using the CGR approach. Our contribution provides a nuanced grading mechanism that accommodates acceptable tolerances in reaction assessments, enhancing the accuracy and flexibility of the grading process.

Pocket Crafter: a 3D generative modeling based workflow for the rapid generation of hit molecules in drug discovery

We present a user-friendly molecular generative pipeline called Pocket Crafter, specifically designed to facilitate hit finding activity in the drug discovery process. This workflow utilized a three-dimensional (3D) generative modeling method Pocket2Mol, for the de novo design of molecules in spatial perspective for the targeted protein structures, followed by filters for chemical-physical properties and drug-likeness, structure-activity relationship analysis, and clustering to generate top virtual hit scaffolds. In our WDR5 case study, we acquired a focused set of 2029 compounds after a targeted searching within Novartis archived library based on the virtual scaffolds. Subsequently, we experimentally profiled these compounds, resulting in a novel chemical scaffold series that demonstrated activity in biochemical and biophysical assays. Pocket Crafter successfully prototyped an effective end-to-end 3D generative chemistry-based workflow for the exploration of new chemical scaffolds, which represents a promising approach in early drug discovery for hit identification.

Integrating QSAR modelling and deep learning in drug discovery: the emergence of deep QSAR

Quantitative structure-activity relationship (QSAR) modelling, an approach that was introduced 60 years ago, is widely used in computer-aided drug design. In recent years, progress in artificial intelligence techniques, such as deep learning, the rapid growth of databases of molecules for virtual screening and dramatic improvements in computational power have supported the emergence of a new field of QSAR applications that we term 'deep QSAR'. Marking a decade from the pioneering applications of deep QSAR to tasks involved in small-molecule drug discovery, we herein describe key advances in the field, including deep generative and reinforcement learning approaches in molecular design, deep learning models for synthetic planning and the application of deep QSAR models in structure-based virtual screening. We also reflect on the emergence of quantum computing, which promises to further accelerate deep QSAR applications and the need for open-source and democratized resources to support computer-aided drug design.

Chemical language models for molecular design

In drug discovery, chemical language models (CLMs) originating from natural language processing offer new opportunities for molecular design. CLMs have been developed using recurrent neural network (RNN) or transformer architectures. For the predictive performance of RNN-based encoder-decoder frameworks and transformers, attention mechanisms play a central role. Among others, emerging application areas for CLMs include constrained generative modeling and the prediction of chemical reactions or drug-target interactions. Since CLMs are applicable to any compound or target data that can be presented in a sequential format and tokenized, mappings of different types of sequences can be learned. For example, active compounds can be predicted from protein sequence motifs. Novel off-the-beat-path applications can also be considered. For example, analogue series from medicinal chemistry can be perceived and represented as chemical sequences and extended with new compounds using CLMs. Herein, methodological features of CLMs and different applications are discussed.

Machine Learning Methods for Small Data Challenges in Molecular Science

Small data are often used in scientific and engineering research due to the presence of various constraints, such as time, cost, ethics, privacy, security, and technical limitations in data acquisition. However, big data have been the focus for the past decade, small data and their challenges have received little attention, even though they are technically more severe in machine learning (ML) and deep learning (DL) studies. Overall, the small data challenge is often compounded by issues, such as data diversity, imputation, noise, imbalance, and high-dimensionality. Fortunately, the current big data era is characterized by technological breakthroughs in ML, DL, and artificial intelligence (AI), which enable data-driven scientific discovery, and many advanced ML and DL technologies developed for big data have inadvertently provided solutions for small data problems. As a result, significant progress has been made in ML and DL for small data challenges in the past decade. In this review, we summarize and analyze several emerging potential solutions to small data challenges in molecular science, including chemical and biological sciences. We review both basic machine learning algorithms, such as linear regression, logistic regression (LR), k-nearest neighbor (KNN), support vector machine (SVM), kernel learning (KL), random forest (RF), and gradient boosting trees (GBT), and more advanced techniques, including artificial neural network (ANN), convolutional neural network (CNN), U-Net, graph neural network (GNN), Generative Adversarial Network (GAN), long short-term memory (LSTM), autoencoder, transformer, transfer learning, active learning, graph-based semi-supervised learning, combining deep learning with traditional machine learning, and physical model-based data augmentation. We also briefly discuss the latest advances in these methods. Finally, we conclude the survey with a discussion of promising trends in small data challenges in molecular science.

Reaction mechanism - explored with the unified reaction valley approach

One of the ultimate goals of chemistry is to understand and manipulate chemical reactions, which implies the ability to monitor the reaction and its underlying mechanism at an atomic scale. In this article, we introduce the Unified Reaction Valley Approach (URVA) as a tool for elucidating reaction mechanisms, complementing existing computational procedures. URVA combines the concept of the potential energy surface with vibrational spectroscopy and describes a chemical reaction via the reaction path and the surrounding reaction valley traced out by the reacting species on the potential energy surface on their way from the entrance to the exit channel, where the products are located. The key feature of URVA is the focus on the curving of the reaction path. Moving along the reaction path, any electronic structure change of the reacting species is registered by a change in the normal vibrational modes spanning the reaction valley and their coupling with the path, which recovers the curvature of the reaction path. This leads to a unique curvature profile for each chemical reaction, with curvature minima reflecting minimal change and curvature maxima indicating the location of important chemical events such as bond breaking/formation, charge polarization and transfer, rehybridization, etc. A decomposition of the path curvature into internal coordinate components or other coordinates of relevance for the reaction under consideration, provides comprehensive insight into the origin of the chemical changes taking place. After giving an overview of current experimental and computational efforts to gain insight into the mechanism of a chemical reaction and presenting the theoretical background of URVA, we illustrate how URVA works for three diverse processes, (i) [1,3] hydrogen transfer reactions; (ii) α-keto-amino inhibitor for SARS-CoV-2 Mpro; (iii) Rh-catalyzed cyanation. We hope that this article will inspire our computational colleagues to add URVA to their repertoire and will serve as an incubator for new reaction mechanisms to be studied in collaboration with our experimental experts in the field.

LinChemIn: SynGraph-a data model and a toolkit to analyze and compare synthetic routes

BACKGROUND

The increasing amount of chemical reaction data makes traditional ways to navigate its corpus less effective, while the demand for novel approaches and instruments is rising. Recent data science and machine learning techniques support the development of new ways to extract value from the available reaction data. On the one side, Computer-Aided Synthesis Planning tools can predict synthetic routes in a model-driven approach; on the other side, experimental routes can be extracted from the Network of Organic Chemistry, in which reaction data are linked in a network. In this context, the need to combine, compare and analyze synthetic routes generated by different sources arises naturally.

RESULTS

Here we present LinChemIn, a python toolkit that allows chemoinformatics operations on synthetic routes and reaction networks. Wrapping some third-party packages for handling graph arithmetic and chemoinformatics and implementing new data models and functionalities, LinChemIn allows the interconversion between data formats and data models and enables route-level analysis and operations, including route comparison and descriptors calculation. Object-Oriented Design principles inspire the software architecture, and the modules are structured to maximize code reusability and support code testing and refactoring. The code structure should facilitate external contributions, thus encouraging open and collaborative software development.

CONCLUSIONS

The current version of LinChemIn allows users to combine synthetic routes generated from various tools and analyze them, and constitutes an open and extensible framework capable of incorporating contributions from the community and fostering scientific discussion. Our roadmap envisages the development of sophisticated metrics for routes evaluation, a multi-parameter scoring system, and the implementation of an entire "ecosystem" of functionalities operating on synthetic routes. LinChemIn is freely available at https://github.com/syngenta/linchemin.

Transformer Performance for Chemical Reactions: Analysis of Different Predictive and Evaluation Scenarios

The prediction of chemical reaction pathways has been accelerated by the development of novel machine learning architectures based on the deep learning paradigm. In this context, deep neural networks initially designed for language translation have been used to accurately predict a wide range of chemical reactions. Among models suited for the task of language translation, the recently introduced molecular transformer reached impressive performance in terms of forward-synthesis and retrosynthesis predictions. In this study, we first present an analysis of the performance of transformer models for product, reactant, and reagent prediction tasks under different scenarios of data availability and data augmentation. We find that the impact of data augmentation depends on the prediction task and on the metric used to evaluate the model performance. Second, we probe the contribution of different combinations of input formats, tokenization schemes, and embedding strategies to model performance. We find that less stable input settings generally lead to better performance. Lastly, we validate the superiority of round-trip accuracy over simpler evaluation metrics, such as top-k accuracy, using a committee of human experts and show a strong agreement for predictions that pass the round-trip test. This demonstrates the usefulness of more elaborate metrics in complex predictive scenarios and highlights the limitations of direct comparisons to a predefined database, which may include a limited number of chemical reaction pathways.

Benchmark of general-purpose machine learning-based quantum mechanical method AIQM1 on reaction barrier heights

Artificial intelligence-enhanced quantum mechanical method 1 (AIQM1) is a general-purpose method that was shown to achieve high accuracy for many applications with a speed close to its baseline semiempirical quantum mechanical (SQM) method ODM2*. Here, we evaluate the hitherto unknown performance of out-of-the-box AIQM1 without any refitting for reaction barrier heights on eight datasets, including a total of ∼24 thousand reactions. This evaluation shows that AIQM1's accuracy strongly depends on the type of transition state and ranges from excellent for rotation barriers to poor for, e.g., pericyclic reactions. AIQM1 clearly outperforms its baseline ODM2* method and, even more so, a popular universal potential, ANI-1ccx. Overall, however, AIQM1 accuracy largely remains similar to SQM methods (and B3LYP/6-31G* for most reaction types) suggesting that it is desirable to focus on improving AIQM1 performance for barrier heights in the future. We also show that the built-in uncertainty quantification helps in identifying confident predictions. The accuracy of confident AIQM1 predictions is approaching the level of popular density functional theory methods for most reaction types. Encouragingly, AIQM1 is rather robust for transition state optimizations, even for the type of reactions it struggles with the most. Single-point calculations with high-level methods on AIQM1-optimized geometries can be used to significantly improve barrier heights, which cannot be said for its baseline ODM2* method.

Predictive chemistry: machine learning for reaction deployment, reaction development, and reaction discovery

The field of predictive chemistry relates to the development of models able to describe how molecules interact and react. It encompasses the long-standing task of computer-aided retrosynthesis, but is far more reaching and ambitious in its goals. In this review, we summarize several areas where predictive chemistry models hold the potential to accelerate the deployment, development, and discovery of organic reactions and advance synthetic chemistry.

Machine learning for yield prediction for chemical reactions using in situ sensors

Machine learning models were developed to predict product formation from time-series reaction data for ten Buchwald-Hartwig coupling reactions. The data was provided by DeepMatter and was collected in their DigitalGlassware cloud platform. The reaction probe has 12 sensors to measure properties of interest, including temperature, pressure, and colour. Colour was a good predictor of product formation for this reaction and machine learning models were able to learn which of the properties were important. Predictions for the current product formation (in terms of % yield) had a mean absolute error of 1.2%. For predicting 30, 60 and 120 min ahead the error rose to 3.4, 4.1 and 4.6%, respectively. The work here presents an example into the insight that can be obtained from applying machine learning methods to sensor data in synthetic chemistry.

Generation of novel Diels-Alder reactions using a generative adversarial network

Deep learning has enormous potential in the chemical and pharmaceutical fields, and generative adversarial networks (GANs) in particular have exhibited remarkable performance in the field of molecular generation as generative models. However, their application in the field of organic chemistry has been limited; thus, in this study, we attempt to utilize a GAN as a generative model for the generation of Diels-Alder reactions. A MaskGAN model was trained with 14 092 Diels-Alder reactions, and 1441 novel Diels-Alder reactions were generated. Analysis of the generated reactions indicated that the model learned several reaction rules in-depth. Thus, the MaskGAN model can be used to generate organic reactions and aid chemists in the exploration of novel reactions.

Fully Automated Generation of Prebiotically Relevant Reaction Networks from Optimized Nanoreactor Simulations

The nanoreactor approach first introduced by the group of Martı́nez [Wang et al. Nat. Chem. 2014, 6, 1044-1048] has recently attracted much attention because of its ability to accelerate the discovery of reaction pathways. Here, we provide a comprehensive study of various simulation parameters and present an alternative implementation for the reactivity-enhancing spherical constraint function, as well as for the detection of reaction events. In this context, a fully automated postsimulation evaluation procedure based on RDKit and NetworkX analysis is introduced. The chemical and physical robustness of the procedure is examined by investigating the reactivity of selected homogeneous systems. The optimized procedure is applied at the GFN2-xTB level of theory to a system composed of HCN molecules and argon atoms, acting as a buffer, yielding prebiotically plausible primary and secondary precursors for the synthesis of RNA. Furthermore, the formose reaction network is explored leading to numerous sugar precursors. The discovered compounds reflect experimental findings; however, new synthetic routes and a large collection of exotic, highly reactive molecules are observed, highlighting the predictive power of the nanoreactor approach for unraveling the reactive manifold.

SELFIES and the future of molecular string representations

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: SELF-referencing embedded string (Selfies). Selfies has since simplified and enabled numerous new applications in chemistry. In this perspective, we look to the future and discuss molecular string representations, along with their respective opportunities and challenges. We propose 16 concrete future projects for robust molecular representations. These involve the extension toward new chemical domains, exciting questions at the interface of AI and robust languages, and interpretability for both humans and machines. We hope that these proposals will inspire several follow-up works exploiting the full potential of molecular string representations for the future of AI in chemistry and materials science.

Improving Chemical Reaction Prediction with Unlabeled Data

Predicting products of organic chemical reactions is useful in chemical sciences, especially when one or more reactants are new organics. However, the performance of traditional learning models heavily relies on high-quality labeled data. In this work, to utilize unlabeled data for better prediction performance, we propose a method that combines semi-supervised learning with graph convolutional neural networks for chemical reaction prediction. First, we propose a Mean Teacher Weisfeiler–Lehman Network to find the reaction centers. Then, we construct the candidate product set. Finally, we use an Improved Weisfeiler–Lehman Difference Network to rank candidate products. Experimental results demonstrate that, with 400k labeled data, our framework can improve the top-5 accuracy by 0.7% using 35k unlabeled data. When the proportion of unlabeled data increases, the performance gain can be larger. For example, with 80k labeled data and 35k unlabeled data, the performance gain with our framework can be 1.8%.

From theory to experiment: transformer-based generation enables rapid discovery of novel reactions

Deep learning methods, such as reaction prediction and retrosynthesis analysis, have demonstrated their significance in the chemical field. However, the de novo generation of novel reactions using artificial intelligence technology requires further exploration. Inspired by molecular generation, we proposed a novel task of reaction generation. Herein, Heck reactions were applied to train the transformer model, a state-of-art natural language process model, to generate 4717 reactions after sampling and processing. Then, 2253 novel Heck reactions were confirmed by organizing chemists to judge the generated reactions. More importantly, further organic synthesis experiments were performed to verify the accuracy and feasibility of representative reactions. The total process, from Heck reaction generation to experimental verification, required only 15 days, demonstrating that our model has well-learned reaction rules in-depth and can contribute to novel reaction discovery and chemical space exploration.

Machine Learning Applications for Chemical Reactions

Machine learning (ML) approaches have enabled rapid and efficient molecular property predictions as well as the design of new novel materials. In addition to great success for molecular problems, ML techniques are applied to various chemical reaction problems that require huge costs to solve with the existing experimental and simulation methods. In this review, starting with basic representations of chemical reactions, we summarized recent achievements of ML studies on two different problems; predicting reaction properties and synthetic routes. The various ML models are used to predict physical properties related to chemical reaction properties (e. g. thermodynamic changes, activation barriers, and reaction rates). Furthermore, the predictions of reactivity, self-optimization of reaction, and designing retrosynthetic reaction paths are also tackled by ML approaches. Herein we illustrate various ML strategies utilized in the various context of chemical reaction studies.

Reproducing the invention of a named reaction: zero-shot prediction of unseen chemical reactions

While state-of-art models can predict reactions through the transfer learning of thousands of samples with the same reaction types as those of the reactions to predict, how to prepare such models to predict "unseen" reactions remains an unanswered question. We aimed to study the Transformer model's ability to predict "unseen" reactions through "zero-shot reaction prediction (ZSRP)", a concept derived from zero-shot learning and zero-shot translation. We reproduced the human invention of the Chan-Lam coupling reaction where the inventor was inspired by the Suzuki reaction when improving Barton's bismuth arylation reaction. After being fine-tuned with samples from these two "existing" reactions, the USPTO-trained Transformer could predict "unseen" Chan-Lam coupling reactions with 55.7% top-1 accuracy. Our model could also mimic the later stage of the history of this reaction, where the initial case of this reaction was generalized to more reactants and reagents via "one-shot/few-shot reaction prediction (OSRP/FSRP)" approaches.

Applications of Artificial Intelligence in Drug Design: Opportunities and Challenges

Artificial intelligence (AI) has undergone rapid development in recent years and has been successfully applied to real-world problems such as drug design. In this chapter, we review recent applications of AI to problems in drug design including virtual screening, computer-aided synthesis planning, and de novo molecule generation, with a focus on the limitations of the application of AI therein and opportunities for improvement. Furthermore, we discuss the broader challenges imposed by AI in translating theoretical practice to real-world drug design; including quantifying prediction uncertainty and explaining model behavior.

Exploring Chemical Reaction Space with Reaction Difference Fingerprints and Parametric t-SNE

Humans prefer visual representations for the analysis of large databases. In this work, we suggest a method for the visualization of the chemical reaction space. Our technique uses the t-SNE approach that is parameterized using a deep neural network (parametric t-SNE). We demonstrated that the parametric t-SNE combined with reaction difference fingerprints could provide a tool for the projection of chemical reactions on a low-dimensional manifold for easy exploration of reaction space. We showed that the global reaction landscape projected on a 2D plane corresponds well with the already known reaction types. The application of a pretrained parametric t-SNE model to new reactions allows chemists to study these reactions in a global reaction space. We validated the feasibility of this approach for two commercial drugs, darunavir and montelukast. We believe that our method can help to explore reaction space and will inspire chemists to find new reactions and synthetic ways.

Combining Machine Learning and Computational Chemistry for Predictive Insights Into Chemical Systems

Machine learning models are poised to make a transformative impact on chemical sciences by dramatically accelerating computational algorithms and amplifying insights available from computational chemistry methods. However, achieving this requires a confluence and coaction of expertise in computer science and physical sciences. This Review is written for new and experienced researchers working at the intersection of both fields. We first provide concise tutorials of computational chemistry and machine learning methods, showing how insights involving both can be achieved. We follow with a critical review of noteworthy applications that demonstrate how computational chemistry and machine learning can be used together to provide insightful (and useful) predictions in molecular and materials modeling, retrosyntheses, catalysis, and drug design.

Reaction Data Curation I: Chemical Structures and Transformations Standardization

The quality of experimental data for chemical reactions is a critical consideration for any reaction-driven study. However, the curation of reaction data has not been extensively discussed in the literature so far. Here, we suggest a 4 steps protocol that includes the curation of individual structures (reactants and products), chemical transformations, reaction conditions and endpoints. Its implementation in Python3 using CGRTools toolkit has been used to clean three popular reaction databases Reaxys, USPTO and Pistachio. The curated USPTO database is available in the GitHub repository (Laboratoire-de-Chemoinformatique/Reaction_Data_Cleaning).

Identifying Chemical Reactions and Their Associated Attributes in Patents

Chemical patents are an essential source of information about novel chemicals and chemical reactions. However, with the increasing volume of such patents, mining information about these chemicals and chemical reactions has become a time-intensive and laborious endeavor. In this study, we present a system to extract chemical reaction events from patents automatically. Our approach consists of two steps: 1) named entity recognition (NER)-the automatic identification of chemical reaction parameters from the corresponding text, and 2) event extraction (EE)-the automatic classifying and linking of entities based on their relationships to each other. For our NER system, we evaluate bidirectional long short-term memory (BiLSTM)-based and bidirectional encoder representations from transformer (BERT)-based methods. For our EE system, we evaluate BERT-based, convolutional neural network (CNN)-based, and rule-based methods. We evaluate our NER and EE components independently and as an end-to-end system, reporting the precision, recall, and F ₁ score. Our results show that the BiLSTM-based method performed best at identifying the entities, and the CNN-based method performed best at extracting events.