Article
Chemistry, Multidisciplinary
Zipeng Zhong, Jie Song, Zunlei Feng, Tiantao Liu, Lingxiang Jia, Shaolun Yao, Min Wu, Tingjun Hou, Mingli Song
Summary: In this article, a root-aligned SMILES (R-SMILES) method is proposed for more efficient synthesis prediction. By comparing it with various state-of-the-art baselines, it is demonstrated that the proposed method significantly outperforms them all, showing its superiority.
Article
Chemistry, Medicinal
Yingchao Yan, Yang Zhao, Huifeng Yao, Jie Feng, Li Liang, Weijie Han, Xiaohe Xu, Chengtao Pu, Chengdong Zang, Lingfeng Chen, Yuanyuan Li, Haichun Liu, Tao Lu, Yadong Chen, Yanmin Zhang
Summary: Retrosynthesis prediction is important in organic synthesis and drug discovery. Existing models lack consideration of the effects of byproducts, which this study addresses by proposing a two-stage retrosynthesis prediction framework based on byproducts. The model achieves high prediction accuracy and demonstrates potential for drug discovery.
JOURNAL OF CHEMICAL INFORMATION AND MODELING
(2023)
Article
Computer Science, Artificial Intelligence
Kelong Mao, Xi Xiao, Tingyang Xu, Yu Rong, Junzhou Huang, Peilin Zhao
Summary: With the challenge of retrosynthesis prediction in chemistry, this study formulates it as a machine translation task and introduces a Graph Enhanced Transformer framework that incorporates both sequential and graphical information of molecules for better performance. The proposed framework outperforms the vanilla Transformer model in test accuracy, showing significant improvements in chemically plausible constrains on atom representation learning.
Review
Plant Sciences
Peipei Wang, Ally M. Schumacher, Shin-Han Shiu
Summary: Predicting plant metabolic pathways is crucial for metabolic engineering and the production of plant metabolite-derived medicine. Recent progress has been made in using multi-omics data and computational approaches to predict the pathways, complementing traditional genetic and biochemical approaches.
CURRENT OPINION IN PLANT BIOLOGY
(2022)
Article
Computer Science, Artificial Intelligence
Esben Jannik Bjerrum, Amol Thakkar, Ola Engkvist
Summary: Automated retrosynthetic planning algorithm is an increasingly important research field. By extracting reaction templates from large datasets and training neural network policies to predict template applicability, the accuracy of route prediction can be improved.
MACHINE LEARNING-SCIENCE AND TECHNOLOGY
(2021)
Review
Chemistry, Multidisciplinary
Corina Marilena Cristache, Ioana Tudor, Liliana Moraru, Gheorghe Cristache, Alessandro Lanza, Mihai Burlibasa
Summary: Our review focuses on the digital design of maxillofacial prostheses, exploring methods of data acquisition for facial defects and assessing software for data processing and part design. Digital workflows have been successfully used for manufacturing extraoral and intraoral prostheses. However, the software and interface are costly and typically accessible only to highly skilled professionals. More user-friendly modules are needed as digital approaches to maxillofacial rehabilitation become more common.
APPLIED SCIENCES-BASEL
(2021)
Article
Chemistry, Multidisciplinary
Baiqing Li, Shimin Su, Chan Zhu, Jie Lin, Xinyue Hu, Lebin Su, Zhunzhun Yu, Kuangbiao Liao, Hongming Chen
Summary: In recent years, artificial intelligence (AI) has revolutionized chemical synthesis, but the lack of suitable representation methods for chemical reactions and scarcity of reaction data hinder the wider application of AI in reaction prediction. This study introduces a novel reaction representation called GraphRXN, which uses a universal graph-based neural network framework to encode chemical reactions directly from two-dimensional reaction structures. The GraphRXN model demonstrates comparable or superior performance to other baseline models when evaluated on three publicly available chemical reaction datasets. Wet-lab experiments were conducted to generate reaction data for further evaluation of the effectiveness of GraphRXN. The model achieved decent accuracy (R-2 of 0.712) when built on high-throughput experimentation data, suggesting its potential for deployment in an integrated workflow combining robotics and AI technologies for forward reaction prediction.
JOURNAL OF CHEMINFORMATICS
(2023)
Article
Chemistry, Multidisciplinary
Junren Li, Lei Fang, Jian-Guang Lou
Summary: Retrosynthesis is an important task in organic chemistry. Recent research has focused on data-driven approaches to improve the prediction accuracy. This study introduces RetroRanker, a ranking model based on graph neural networks, to mitigate the frequency bias issue in existing retrosynthesis models through re-ranking. The results on public retrosynthesis benchmarks demonstrate the improvement achieved by RetroRanker on state-of-the-art models, and preliminary studies indicate its potential in enhancing the performance of multi-step retrosynthesis.
JOURNAL OF CHEMINFORMATICS
(2023)
Article
Biotechnology & Applied Microbiology
Venkat Venkatasubramanian, Vipul Mann
Summary: Recent years have witnessed a rapid increase in the use of artificial intelligence methods for computational reaction modeling and prediction. These methods can be broadly categorized into symbolic AI, purely data-driven numeric AI, and hybrid AI. Symbolic AI translates prior chemistry knowledge into rules, numeric AI utilizes machine learning without explicit domain knowledge, and hybrid AI integrates domain knowledge with data-driven techniques.
CURRENT OPINION IN CHEMICAL ENGINEERING
(2022)
Review
Engineering, Multidisciplinary
Yinjie Jiang, Yemin Yu, Ming Kong, Yu Mei, Luotian Yuan, Zhengxing Huang, Kun Kuang, Zhihua Wang, Huaxiu Yao, James Zou, Connor W. Coley, Ying Wei
Summary: In recent years, there has been a significant increase in interest in AI-driven retrosynthesis prediction. This review provides an overview of the current landscape and discusses the challenges and progress in this field. It also proposes a novel framework for categorizing different components of retrosynthesis prediction and examines how AI is reshaping each component.
Article
Chemistry, Medicinal
Haris Hasic, Takashi Ishida
Summary: The application of retrosynthesis requires a lot of chemistry knowledge and experience, but it becomes almost impossible for human chemists to efficiently apply it to compounds with intricate molecular structures. With the advancement of machine learning and artificial intelligence, computer-assisted retrosynthesis has gained research attention again. A novel template-free approach has been developed to address the challenge of low exploration ability in recent retrosynthesis methods.
JOURNAL OF CHEMICAL INFORMATION AND MODELING
(2021)
Article
Engineering, Chemical
Kevin Zhang, Vipul Mann, Venkat Venkatasubramanian
Summary: This study proposes a novel chemistry-aware retrosynthesis prediction framework that combines data-driven models with prior domain knowledge, achieving significant performance improvements.
Article
Chemistry, Multidisciplinary
Umit V. Ucak, Taek Kang, Junsu Ko, Juyong Lee
Summary: In this study, the retrosynthetic planning problem is redefined as a language translation issue, using a template-free sequence-to-sequence model that represents chemical reactions based on molecular fragments. This new approach demonstrates better prediction results than current state-of-the-art computational methods and resolves major shortcomings of existing methods.
JOURNAL OF CHEMINFORMATICS
(2021)
Article
Computer Science, Theory & Methods
Zari Shamsa, Ali Rezaee, Sahar Adabi, Amir Masoud Rahmani
Summary: The processing of data from new communication devices, such as IoT-based technology, has significantly increased in the past decade. Resource management is crucial for improving efficiency in cloud/fog-based platforms. However, most existing methods may not achieve optimal load balancing, affecting quality of service and customer satisfaction. This paper proposes a 4-layer software architecture for analyzing workflows and dynamic resources in cloud/fog/IoT environments, which considers workload and presence prediction of IoT nodes. It also suggests architecture components to meet quality attributes and evaluates the proposed architecture using the ATAM method.
Article
Biochemical Research Methods
Han Bao, Jinhui Zhao, Xinjie Zhao, Chunxia Zhao, Xin Lu, Guowang Xu
Summary: In this study, we proposed a transfer learning approach using a pre-trained hybrid deep learning architecture, GTC, which combines Graph Transformer and convolutional neural network, to predict plant metabolic pathways. GTC provides comprehensive molecular representation and outperforms other machine learning models on the KEGG dataset. It achieves high accuracy in classifying KEGG metabolic pathways and predicting plant secondary metabolic pathways. Furthermore, GTC demonstrates its generalization ability by accurately classifying natural products.
BMC BIOINFORMATICS
(2023)
Article
Biotechnology & Applied Microbiology
Sara Castano-Cerezo, Mathieu Fournie, Philippe Urban, Jean-Loup Faulon, Gilles Truan
FRONTIERS IN BIOENGINEERING AND BIOTECHNOLOGY
(2020)
Review
Audiology & Speech-Language Pathology
Julia Sidorova, Pablo Carbonell, Milena Cukic
Summary: The possibility of estimating glucose value from voice has the potential to revolutionize diabetes treatment by providing nonintrusive and instantaneous blood glucose estimation. This review provides a comprehensive overview of different approaches in this field and includes a transparency and reproducibility score to assess the biases in primary research. The discussion on future research pathways further highlights the importance of this topic.
Article
Biotechnology & Applied Microbiology
Irene Otero-Muras, Pablo Carbonell
Summary: Metabolic engineering involves optimizing processes from single-cell to fermentation to increase production of valuable chemicals. A systems approach has accelerated scaling from rapid prototyping to industrial production, with automated DNA assembly reducing time from conception to production. The success of metabolic engineering often relies on retrobiosynthetic protocols and dynamic regulation strategies assembled as genetic circuits in host strains.
METABOLIC ENGINEERING
(2021)
Review
Biochemistry & Molecular Biology
Christopher J. Robinson, Jonathan Tellechea-Luzardo, Pablo Carbonell, Adrian J. Jervis, Cunyu Yan, Katherine A. Hollywood, Mark S. Dunstan, Andrew Currin, Eriko Takano, Nigel S. Scrutton
Summary: Metabolic engineering technologies have been successfully employed for the engineering and optimization of industrial host strains over the past three decades. Design-Build-Test-Learn pipelines are being established to rapidly deliver diverse chemical targets through iterative optimization of microbial production strains. Biofoundries are using in silico tools for genetic design and combinatorial design of experiments to optimize selection within the potential design space based on multi-criteria objectives.
BIOCHEMICAL SOCIETY TRANSACTIONS
(2021)
Article
Microbiology
Pascal Hilaire, Sandy Contreras, Helene Blanquart-Goudezeune, Jonathan Verbeke, Baudoin Delepine, Lucas Marmiesse, Remi Peyraud, Stanislas C. Morand
Summary: The complete genome sequence of Sphingobium xenophagum strain PH3-15, isolated from La Roche-Posay thermal water sources, consists of two chromosomes and three plasmids totaling 4.6 Mbp. These findings provide valuable information and insights into the physiology and metabolism of this particular Sphingobium organism.
MICROBIOLOGY RESOURCE ANNOUNCEMENTS
(2021)
Review
Biotechnology & Applied Microbiology
Jonathan Tellechea-Luzardo, Irene Otero-Muras, Angel Goni-Moreno, Pablo Carbonell
Summary: Biofoundries are highly automated facilities that enable rapid and efficient processes in biomanufacturing and engineering biology. However, they can be costly and time-consuming to develop. By considering strategies early on and identifying bottlenecks, biofoundries can be optimized for efficiency. This article provides a survey of technological solutions and explores pathways towards the creation of a functional biofoundry.
TRENDS IN BIOTECHNOLOGY
(2022)
Article
Biochemical Research Methods
Angelo Cardoso Batista, Antoine Levrier, Paul Soudier, Peter L. Voyvodic, Tatjana Achmedov, Tristan Reif-Trauttmansdor, Angelique DeVisch, Martin Cohen-Gonsaud, Jean-Loup Faulon, Chase L. Beisel, Jerome Bonnet, Manish Kushwaha
Summary: This study presents a simple, efficient, and cost-effective solution for using linear DNA templates in cell-free systems by deleting the exonuclease gene cluster from Escherichia coli. The research highlights the importance of tailoring buffer composition for the optimal experimental setup, and suggests that similar strategies can be applied to other species in cell-free synthetic biology.
ACS SYNTHETIC BIOLOGY
(2022)
Article
Biochemical Research Methods
Paul Soudier, Ana Zuniga, Thomas Duigou, Peter L. Voyvodic, Kenza Bazi-Kabbaj, Manish Kushwaha, Julie A. Vendrell, Jerome Solassol, Jerome Bonnet, Jean-Loup Faulon
Summary: This study reports the engineering of PeroxiHUB, a sensing platform centered around H2O2, that supports cell-free detection of different metabolites. The PeroxiHUB platform utilizes enzymatic transducers to convert metabolites of interest into H2O2, allowing for rapid reprogramming of sensor specificity. This platform has the potential to detect a wide range of metabolites in a modular and scalable fashion.
ACS SYNTHETIC BIOLOGY
(2022)
Article
Multidisciplinary Sciences
Joan Herisson, Thomas Duigou, Melchior du Lac, Kenza Bazi-Kabbaj, Mahnaz Sabeti Azad, Gizem Buldum, Olivier Telle, Yorgo El Moubayed, Pablo Carbonell, Neil Swainston, Valentin Zulkower, Manish Kushwaha, Geoff S. Baldwin, Jean-Loup Faulon
Summary: Introduced the Galaxy-SynBioCAD portal, a toolshed for synthetic biology, metabolic engineering, and industrial biotechnology. The portal provides tools and workflows to facilitate the design and engineering process of metabolic pathways, from strain and target selection to DNA part design and plasmid assembly.
NATURE COMMUNICATIONS
(2022)
Review
Biotechnology & Applied Microbiology
Jonathan Tellechea-Luzardo, Martin T. Stiebritz, Pablo Carbonell
Summary: Advances in synthetic biology and genetic engineering have led to the development of transcription factor (TF)-based biosensors, which are promising tools for detecting chemical compounds and eliciting specific responses. However, widespread use of these biosensors is hindered by challenges that can be addressed by increasing knowledge of metabolite-activated transcription factors, identifying new transcription factors, and improving the design workflow for biosensor circuits. These improvements are especially important in the bioproduction field, where better biosensor-based approaches are needed for screening and regulation. This work summarizes the current understanding of TF-based biosensors, discusses recent experimental and computational approaches for modification and improvement, and suggests future research directions for bioproduction screening and genetic circuit regulation.
FRONTIERS IN BIOENGINEERING AND BIOTECHNOLOGY
(2023)
Article
Biochemical Research Methods
Jonathan Tellechea-Luzardo, Hector Martin Lazaro, Raul Moreno Lopez, Pablo Carbonell
Summary: Allosteric transcription factor (aTF) based biosensors are widely used in engineering genetic circuits. However, the scattered and incomplete knowledge about validated molecule-TF pairs and the limited number of known TF-compound interactions pose challenges in finding new TF-ligand pairs. In this study, we present Sensbio, a computational tool that utilizes a TF-ligand reference database to identify potential transcription factors activated by specific input molecules. Our tool includes algorithms, an online application, and a predictive model based on machine learning for discovering new matches.
BMC BIOINFORMATICS
(2023)
Article
Multidisciplinary Sciences
Leon Faure, Bastien Mollet, Wolfram Liebermeister, Jean-Loup Faulon
Summary: Constraint-based metabolic models have been used to predict microorganism phenotype, but accurate predictions require labor-intensive measurements. We propose hybrid neural-mechanistic models as a machine learning architecture to improve phenotype predictions. Our models outperform constraint-based models with smaller training set sizes, offering a time and resource-saving approach in systems biology and biological engineering projects.
NATURE COMMUNICATIONS
(2023)
Article
Biochemical Research Methods
Jing Wui Yeoh, Neil Swainston, Peter Vegh, Valentin Zulkower, Pablo Carbonell, Maciej B. Holowko, Gopal Peddinti, Chueh Loo Poh
Summary: Advances in hardware automation in synthetic biology laboratories have outpaced those in software development. The development of software solutions is crucial for automated laboratories to assist with specialized tasks. Many labs are independently developing similar software solutions, highlighting the need for standardized open-source software packages.
Article
Biotechnology & Applied Microbiology
Nicolas Huber, Edgar Alberto Alcala-Orozco, Thomas Rexer, Udo Reichl, Steffen Klamt
Summary: Cell-free production systems are commonly used for synthesizing industrial chemicals and biopharmaceuticals. This study presents a model-based optimization framework for cell-free enzyme cascades, taking into account parameter uncertainties. The approach was exemplified using the synthesis of GDP-fucose, resulting in significant improvements in the process.
METABOLIC ENGINEERING
(2024)