Enzyme catalysts are an integral part of green chemistry strategies towards a more sustainable and resource-efficient chemical synthesis. However, the use of biocatalysed reactions in retrosynthetic planning clashes with the difficulties in predicting the enzymatic activity on unreported substrates and enzyme-specific stereo- and regioselectivity. As of now, only rule-based systems support retrosynthetic planning using biocatalysis, while initial data-driven approaches are limited to forward predictions. Here, we extend the data-driven forward reaction as well as retrosynthetic pathway prediction models based on the Molecular Transformer architecture to biocatalysis. The enzymatic knowledge is learned from an extensive data set of publicly available biochemical reactions with the aid of a new class token scheme based on the enzyme commission classification number, which captures catalysis patterns among different enzymes belonging to the same hierarchy. The forward reaction prediction model (top-1 accuracy of 49.6%), the retrosynthetic pathway (top-1 single-step round-trip accuracy of 39.6%) and the curated data set are made publicly available to facilitate the adoption of enzymatic catalysis in the design of greener chemistry processes.

With the growing availability of data within various scientific domains, generative models hold enormous potential to accelerate scientific discovery at every step of the scientific method. Perhaps their most valuable application lies in the speeding up of what has traditionally been the slowest and most challenging step of coming up with a hypothesis. Powerful representations are now being learned from large volumes of data to generate novel hypotheses, which is making a big impact on scientific discovery applications ranging from material design to drug discovery. The GT4SD [Team, 2022] (https://github.com/GT4SD/gt4sd-core) is an extensible open-source library that enables scientists, developers and researchers to train and use state-of-the-art generative models for hypothesis generation in scientific discovery. GT4SD supports a variety of uses of generative models across material science and drug discovery, including molecule discovery and design based on properties related to target proteins, omic profiles, scaffold distances, binding energies and more.Keywords Generative Models • Scientific Discovery • Accelerated Discovery • Open Source Humanity's progress has been characterised by a delicate balance between curiosity and creativity. Science is no exception with its long evolution through trial and error. While remarkably successful, the scientific method can be a slow iterative process that can be inadequate when faced with important and pressing needs, e.g., the need to swiftly develop drugs and antibiotics or design novel materials and processes to mitigate climate change effects. Indeed, it can take almost a decade to discover a new material and cost upwards of $10-$100 million. One of the most daunting challenges in materials discovery is hypothesis generation, where it is extremely challenging to identify and select novel and useful candidates in search spaces that are overwhelming in size, e.g., the chemical space for drug-like molecules is estimated to contain 10 33 structures [Polishchuk et al., 2013].To overcome this problem, in recent years, generative models have emerged as an effective approach to design and discover molecules with desired properties. Generative models more efficiently and effectively navigate and explore vast search spaces that are learned from data based on user-defined criteria. Starting from a series of seminal works [Gómez-

Recent advances in language modeling have tremendously impacted how we handle sequential data in science. Language architectures have emerged as a hotbed of innovation and creativity in natural language processing over the last decade, and have since gained prominence in modeling proteins and chemical processes, elucidating structural relationships from textual/sequential data. Surprisingly, some of these relationships refer to three-dimensional structural features, raising important questions on the dimensionality of the information contained in sequential data. We demonstrate that the unsupervised use of a language model architecture to a language representation of bio-catalyzed chemical reactions can capture the signal at the base of the substrate-active site atomic interactions, identifying the three- dimensional active site position in unknown protein sequences. The language representation comprises a reaction-simplified molecular-input line-entry system (SMILES) for substrate and products, and amino acid sequence information for the enzyme. This approach can recover, with no supervision, 52.12% of the active site when considering co-crystallized substrate-enzyme structures as ground truth, vastly outperforming other attention-based models.

With the growing availability of data within various scientific domains, generative models hold enormous potential to accelerate scientific discovery. They harness powerful representations learned from datasets to speed up the formulation of novel hypotheses with the potential to impact material discovery broadly. We present the Generative Toolkit for Scientific Discovery (GT4SD). This extensible open-source library enables scientists, developers, and researchers to train and use state-of-the-art generative models to accelerate scientific discovery focused on organic material design.

The first decade of genome sequencing saw a surge in the characterization of proteins with unknown functionality. Even still, more than 20% of proteins in well-studied model animals have yet to be identified, making the discovery of their active site one of biology's greatest puzzle. Herein, we apply a Transformer architecture to a language representation of bio-catalyzed chemical reactions to learn the signal at the base of the substrate-active site atomic interactions. The language representation comprises a reaction simplified molecular-input line-entry system (SMILES) for substrate and products, complemented with amino acid (AA) sequence information for the enzyme. We demonstrate that by creating a custom tokenizer and a score based on attention values, we can capture the substrate-active site interaction signal and utilize it to determine the active site position in unknown protein sequences, unraveling complicated 3D interactions using just 1D representations. This approach exhibits remarkable results and can recover, with no supervision, 31.51% of the active site when considering co-crystallized substrate-enzyme structures as a ground-truth, vastly outperforming approaches based on sequence similarities only. Our findings are further corroborated by docking simulations on the 3D structure of few enzymes. This work confirms the unprecedented impact of natural language processing and more specifically of the Transformer architecture on domain-specific languages, paving the way to effective solutions for protein functional characterization and bio-catalysis engineering.

Enzyme catalysts are an integral part of green chemistry strategies towards a more sustainable and resource-efficient chemical synthesis. However, the use of enzymes on unreported substrates and their specific stereo- and regioselectivity are domain-specific knowledge factors that require decades of field experience to master. This makes the retrosynthesis of given targets with biocatalysed reactions a significant challenge. Here, we use the molecular transformer architecture to capture the latent knowledge about enzymatic activity from a large data set of publicly available biochemical reactions, extending forward reaction and retrosynthetic pathway prediction to the domain of biocatalysis. We introduce the use of a class token based on the EC classification scheme that allows to capture catalysis patterns among different enzymes belonging to the same hierarchical families. The forward prediction model achieves an accuracy of 49.6% and 62.7%, top-1 and top-5 respectively, while the single-step retrosynthetic model shows a round-trip accuracy of 39.6% and 42.6%, top-1 and top-10 respectively. Trained models and curated data are made publicly available with the hope of promoting enzymatic catalysis and making green chemistry more accessible through the use of digital technologies.

The first decade of genome sequencing saw a surge in the characterization of proteins with unknown functionality. Even still, more than 20% of proteins in well-studied model animals have yet to be identified, making the discovery of their active site one of biology's greatest puzzle. Herein, we apply a Transformer architecture to a language representation of bio-catalyzed chemical reactions to learn the signal at the base of the substrate-active site atomic interactions. The language representation comprises a reaction simplified molecular-input line-entry system (SMILES) for substrate and products, complemented with amino acid (AA) sequence information for the enzyme. We demonstrate that by creating a custom tokenizer and a score based on attention values, we can capture the substrate-active site interaction signal and utilize it to determine the active site position in unknown protein sequences, unraveling complicated 3D interactions using just 1D representations. This approach exhibits remarkable results and can recover, with no supervision, 31.51% of the active site when considering co-crystallized substrate-enzyme structures as a ground-truth, vastly outperforming approaches based on sequence similarities only. Our findings are further corroborated by docking simulations on the 3D structure of few enzymes. This work confirms the unprecedented impact of natural language processing and more specifically of the Transformer architecture on domain-specific languages, paving the way to effective solutions for protein functional characterization and bio-catalysis engineering.

The first decade of genome sequencing saw a surge in the characterization of proteins with unknown functionality. Even still, more than 20% of proteins in well-studied model animals have yet to be identified, making the discovery of their active site one of biology's greatest puzzle. Herein, we apply a Transformer architecture to a language representation of bio-catalyzed chemical reactions to learn the signal at the base of the substrate-active site atomic interactions. The language representation comprises a reaction simplified molecular-input line-entry system (SMILES) for substrate and products, complemented with amino acid (AA) sequence information for the enzyme. We demonstrate that by creating a custom tokenizer and a score based on attention values, we can capture the substrate-active site interaction signal and utilize it to determine the active site position in unknown protein sequences, unraveling complicated 3D interactions using just 1D representations. This approach exhibits remarkable results and can recover, with no supervision, 31.51% of the active site when considering co-crystallized substrate-enzyme structures as a ground-truth, vastly outperforming approaches based on sequence similarities only. Our findings are further corroborated by docking simulations on the 3D structure of few enzymes. This work confirms the unprecedented impact of natural language processing and more specifically of the Transformer architecture on domain-specific languages, paving the way to effective solutions for protein functional characterization and bio-catalysis engineering.