Review
Chemistry, Physical
Kamil Paduszynski, Krzysztof Klebowski, Marta Krolikowska
Summary: The article reviews quantitative structure-property relationships for predicting melting point temperature of ionic liquids and proposes new models using experimental data for 953 salts. A variety of machine learning algorithms are applied, including regression and classification methods.
JOURNAL OF MOLECULAR LIQUIDS
(2021)
Article
Chemistry, Multidisciplinary
Medard Edmund Mswahili, Min-Jeong Lee, Gati Lother Martin, Junghyun Kim, Paul Kim, Guang J. Choi, Young-Seob Jeong
Summary: The study found that artificial neural networks show great potential in predicting cocrystal formation, exhibiting the best accuracy, sensitivity, and F1 score. This has significant implications for accelerating and improving cocrystal development.
APPLIED SCIENCES-BASEL
(2021)
Article
Pharmacology & Pharmacy
Jie Liu, Wenjing Guo, Fan Dong, Jason Aungst, Suzanne Fitzpatrick, Tucker A. A. Patterson, Huixiao Hong
Summary: This study demonstrates the potential of using machine learning models to evaluate the reproductive toxicity of chemicals. By developing predictive models based on rat multigeneration reproductive toxicity testing data and seven machine learning algorithms, researchers were able to achieve good performance in predicting reproductive toxicity. The findings suggest that machine learning can be a promising alternative approach in assessing the potential reproductive toxicity of chemicals.
FRONTIERS IN PHARMACOLOGY
(2022)
Article
Chemistry, Medicinal
Michael R. Maser, Alexander Y. Cui, Serim Ryou, Travis J. DeLano, Yisong Yue, Sarah E. Reisman
Summary: Machine-learned ranking models have been developed for predicting substrate-specific cross-coupling reaction conditions. Graph encodings and gradient-boosting machines were found to be very effective for this learning task, with a novel reaction-level graph attention operation in the top-performing model.
JOURNAL OF CHEMICAL INFORMATION AND MODELING
(2021)
Article
Engineering, Chemical
Zihao Wang, Zhen Song, Teng Zhou
Summary: Machine learning models were developed to predict the toxicity of ionic liquids, with the support vector machine algorithm slightly outperforming the feedforward neural network algorithm; after structure optimization through five-fold cross validation, the models exhibited high predictive accuracy and can be useful for computer-aided molecular design of environmentally friendly ILs.
Article
Engineering, Chemical
Tianxiong Liu, Dingchao Fan, Yusen Chen, Yasen Dai, Yuyang Jiao, Peizhe Cui, Yinglong Wang, Zhaoyou Zhu
Summary: In this study, a novel molecular structure encoding method was developed, and a convolutional autoencoder was used for denoising based on the structure of ionic liquids (ILs). Combined with three different machine learning algorithms (multilayer perceptron [MLP], random forest [RF], and support vector machine [SVM]), it showed better prediction performance for CO2 solubility in ILs than conventional cheminformatics descriptors. The SE-MLP model achieved the best performance with an R-2 value of 0.9873 and a mean square error of 0.0007.
Article
Chemistry, Multidisciplinary
Anastasia K. Lavrinenko, Ivan Yu. Chernyshov, Evgeny A. Pidko
Summary: This study presents an alternative model based on machine learning for predicting the melting temperatures of binary metal-free DESs or ionic liquids. The model integrates experimental data, computational simulations, and cheminformatic descriptors, and demonstrates high accuracy in a large-scale database.
ACS SUSTAINABLE CHEMISTRY & ENGINEERING
(2023)
Article
Chemistry, Physical
Yi Ding, Minchun Chen, Chao Guo, Peng Zhang, Jingwen Wang
Summary: This study developed quantitative structure-activity relationship (QSAR) models for predicting the properties of ionic liquids (ILs) by using molecular fingerprint (MF) and machine learning (ML). The models showed comparative predictive performance with traditional methods and can quickly obtain representations of IL within milliseconds. The trustworthiness of the models was enhanced by interpreting them using the shapely additive explanation (SHAP) method.
JOURNAL OF MOLECULAR LIQUIDS
(2021)
Review
Pharmacology & Pharmacy
Masataka Kuroda, Reiko Watanabe, Tsuyoshi Esaki, Hitoshi Kawashima, Rikiya Ohashi, Tomohiro Sato, Teruki Honma, Hiroshi Komura, Kenji Mizuguchi
Summary: This article discusses one solution to address the shortage of publicly available data, which is to collect more quality-controlled data from the private sector through public-private partnerships. Using a case study in Japan, the technical aspects of these partnerships are reviewed, with a focus on data collection from multiple private sector companies and its impact on improving chemical space coverage and prediction performance when merged with public sector datasets.
DRUG DISCOVERY TODAY
(2022)
Article
Automation & Control Systems
Dingling Kong, Yue Luan, Xiaowei Zhao, Yanhua Lu, Wei Li, Qingyou Zhang, Aimin Pang
Summary: This study collected 17817 compounds and used various methods to predict their melting points. The random forest method achieved the best results and outperformed previous literature. Furthermore, combining the descriptor suggested in this study with other descriptors can further improve the prediction accuracy.
CHEMOMETRICS AND INTELLIGENT LABORATORY SYSTEMS
(2023)
Article
Thermodynamics
Pratik Dhakal, Jindal K. Shah
Summary: Two machine learning models, support vector machine and artificial neural network, were developed to predict ionic conductivity of pure ionic liquids. The artificial neural network model accurately predicted the conductivity for 1102 ionic liquids, including mixtures exhibiting nonideal behavior.
FLUID PHASE EQUILIBRIA
(2021)
Article
Chemistry, Physical
D. M. Makarov, Yu A. Fadeeva, L. E. Shmukler, I. Tetko
Summary: In this study, a rigorous validation protocol was used to predict the melting point of ionic liquids (ILs), with a recommendation to use this protocol for validating models of other IL properties. The Transformer Convolutional Neural Network model based on text representation showed significantly higher prediction accuracy than models using traditional descriptors.
JOURNAL OF MOLECULAR LIQUIDS
(2021)
Article
Chemistry, Physical
Affaf Djihed Boualem, Kadda Argoub, Ali Mustapha Benkouider, Ahmed Yahiaoui, Khaled Toubal
Summary: In this study, two alternative models for predicting the dynamic viscosity of ionic liquids as a function of temperature were proposed using molecular functional groups. The models were developed through regression analysis and machine learning techniques, and tested using a large experimental database. The results showed that the support vector machine regression model outperformed the traditional non-linear model in terms of accuracy and reliability.
JOURNAL OF MOLECULAR LIQUIDS
(2022)
Article
Multidisciplinary Sciences
Qi-Jun Hong, Sergey V. Ushakov, Axel van de Walle, Alexandra Navrotsky
Summary: In this study, a machine learning model was developed to predict the melting temperature of compounds quickly, overcoming the time-consuming nature of traditional measurements or computations. The model has demonstrated its usefulness in various fields, such as materials design and discovery, as well as planetary science and geology.
PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA
(2022)
Article
Chemistry, Physical
Jingzi Zhang, Zhuoxuan Zhu, X. -D. Xiang, Ke Zhang, Shangchao Huang, Chengquan Zhong, Hua-Jun Qiu, Kailong Hu, Xi Lin
Summary: This study introduces a structural descriptor called smooth overlap of atomic position (SOAP) into machine learning (ML) models to predict the T-c values of superconductors. The ML models achieved a high prediction accuracy using the SOAP descriptor and identified new high-temperature superconductors.
JOURNAL OF PHYSICAL CHEMISTRY C
(2022)
Article
Chemistry, Physical
Kaycee Low, Michelle L. Coote, Ekaterina I. Izgorodina
Summary: By incorporating electronic information into the descriptor, QM descriptors improve the accuracy and data efficiency of machine learning predictions for interaction energy. The EDDIE-ML model performs well under low-data conditions and can be easily transferred to more complex systems.
JOURNAL OF CHEMICAL THEORY AND COMPUTATION
(2022)
Article
Chemistry, Multidisciplinary
Anh L. P. Nguyen, Ekaterina I. Izgorodina
Summary: This study investigated the performance of the counterpoise (CP) correction by Boys and Bernardi in many-body clusters of organic compounds. It was found that the CP-corrected interaction energies were basis-set independent, unlike the electronic energies of individual molecules. The use of a small basis set showed excellent performance in predicting HF interaction energies.
JOURNAL OF COMPUTATIONAL CHEMISTRY
(2022)
Article
Chemistry, Multidisciplinary
Zoe L. Seeger, Ekaterina I. Izgorodina
Summary: The CCSD(T) method is considered the gold standard of computational chemistry for its accurate recovery of correlation energy, but its application has been limited to medium-sized molecular systems. The development of the DLPNO-CCSD(T) method has significantly broadened the range of chemical systems to which CCSD(T) level calculations can be applied.
JOURNAL OF COMPUTATIONAL CHEMISTRY
(2022)
Article
Chemistry, Multidisciplinary
Musen Li, Rika Kobayashi, Roger D. Amos, Michael J. Ford, Jeffrey R. Reimers
Summary: The correction of the asymptotic potential error in density functionals improves the accuracy of calculated molecular excited states with charge-transfer character. New computational methods in materials science demonstrate these effects in materials spectroscopy. Defects play a bridging role between molecular and materials spectroscopies, enhancing the accuracy of calculations in these areas.
Editorial Material
Astronomy & Astrophysics
Abigail J. Frost, Ryan M. Lau, Leonard Burtscher, Chris Packham, Elizabeth J. Tasker, Glen A. Rees, Vanessa A. Moss, Rika Kobayashi
Summary: This article discusses the convening of the IR 2022 conference shortly after the launch of the James Webb Space Telescope, aiming to identify and explore synergies between ground-based and space-based infrared observations.
Article
Chemistry, Medicinal
Kaycee Low, Michelle L. Coote, Ekaterina I. Izgorodina
Summary: This study proposes a graph neural network model for predicting the solvation Gibbs free energy of molecules. By incorporating chemically intuitive solvation-relevant parameters into the featurization process, the model accurately predicts solvation energy. In testing, the model shows comparable accuracy to traditional methods and provides clear explanations.
JOURNAL OF CHEMICAL INFORMATION AND MODELING
(2022)
Article
Polymer Science
Abhishek Singh, Thomas G. Mason, Zhenzhen Lu, Boon Mia Teo, Benny D. Freeman, Ekaterina I. Izgorodina
Summary: Mussel-inspired polycatecholamine coatings and films are versatile materials for surface modification. However, the progress of chemically modified coatings with tailored properties has been slow due to the lack of definitive evidence of their polymeric structure. This study reports the first solution-phase polymerization approach using ionic liquids to achieve polymerization of four catecholamines, resulting in complex heterogeneous and soluble polymers.
MACROMOLECULAR CHEMISTRY AND PHYSICS
(2022)
Article
Polymer Science
Thomas G. Mason, Benny D. Freeman, Ekaterina I. Izgorodina
Summary: This study introduces a new polymer formation scheme that takes into account the propagation ratios of two monomers. The scheme was implemented as an extension of the Polymatic code, and it was found that influencing the polymerization process can enhance the tensile strength of the membranes. However, it does not necessarily improve the performance of the membranes. Quantum calculations should be performed before molecular dynamics simulations to determine the need for intramolecular cross-linking.
Article
Chemistry, Physical
Kaycee Low, Michelle L. Coote, Ekaterina I. Izgorodina
Summary: This work extends the electron deformation density-based descriptor to predict three-body interactions in trimers. The resulting Gaussian process regression (GPR) model shows good prediction accuracy on the 3B69 and S22-3 trimer data sets. A hybrid kernel function is introduced to predict the total trimer interaction energy and three-body contribution using the same descriptor. A new data set based on protein-ligand crystal structures is introduced, providing benchmark calculations for larger molecular interactions. Compared to DFT- and wavefunction-based methods, our model reduces the required SCF calculations and shows faster predictions.
JOURNAL OF CHEMICAL THEORY AND COMPUTATION
(2023)
Article
Chemistry, Physical
Roger D. D. Amos, Rika Kobayashi
Summary: The recent discovery of interstellar molecules HCCCHCCC and HCCCCS in TMC-1 has prompted an ab initio investigation of these isomers. While the isomers of HCCCHCCC have been previously studied, the research on HCCCCS is limited. In this paper, high-level CCSD(T) calculations are performed to study the lowest eleven isomers of HCCCCS, and complete basis set extrapolation is applied to further investigate the key isomers and provide high-precision rotational constants that could assist astronomers in their assignments.
Article
Chemistry, Multidisciplinary
Marco Fronzi, Roger D. Amos, Rika Kobayashi
Summary: We analysed the efficacy of machine learning interatomic potentials in modelling gold nanoparticles. We explored their transferability to larger systems and established simulation times and size thresholds for accurate potentials. By comparing the energies and geometries of large gold nanoclusters, we gained insight into the number of simulation timesteps required to generate ML-IPs that replicate structural properties. Our findings suggest that minor adjustments to potentials can make them suitable for different systems. These results contribute to the development of accurate interatomic potentials for modelling gold nanoparticles using machine learning techniques.
Editorial Material
Astronomy & Astrophysics
Sarah M. Wagner, Beatriz Mingo, Fatemeh Zahra Majidi, Andrea Gokus, Leonard Burtscher, Cenk Kayhan, Rika Kobayashi, Parit Mehta, Vanessa A. Moss, Volker Ossenkopf-Okada, Ken Rice, Adam R. H. Stevens, Gaurav Waratkar, Paul Woods
Summary: More than 500 participants from around the world attended the first Astronomers for Planet Earth Symposium, aiming to discuss and promote sustainability in the field of astronomy, as well as explore opportunities for astronomers to contribute to climate communication.
Article
Chemistry, Multidisciplinary
Amy L. Thomson, Ellen C. Gleeson, Alessia Belgi, W. Roy Jackson, Ekaterina I. Izgorodina, Andrea J. Robinson
Summary: Ru-Alkylidene catalysed olefin metathesis can generate metabolically stable peptidomimetics with defined geometry of cystine bridge. By in situ and reversible oxidation of thiol and thioether functionality to disulfides and S-oxides, respectively, the deleterious coordinative bonding between sulfur-containing functionality found in cysteine and methionine residues and the catalyst can be negated, facilitating high yielding ring-closing and cross metathesis of bioorthogonally protected peptides.
CHEMICAL COMMUNICATIONS
(2023)
Article
Chemistry, Physical
Fairuz H. H. Hashim, Fiona Yu, Ekaterina I. I. Izgorodina
Summary: Thermodynamics studies how energy is stored, transformed and transferred, but predicting properties with simulation techniques is challenging due to several factors. In this study, QCE theory is applied to predict the thermodynamic properties of liquid water by representing it as clusters. The appropriate selection of clusters is crucial for accurate predictions.
PHYSICAL CHEMISTRY CHEMICAL PHYSICS
(2023)
Article
Chemistry, Physical
Abhishek Singh, Thomas G. Mason, Zhenzhen Lu, Anita J. Hill, Steven J. Pas, Boon Mia Teo, Benny D. Freeman, Ekaterina I. Izgorodina
Summary: Minimal understanding of the formation mechanism and structure of polydopamine (pDA) and its natural analogue, eumelanin, hinders their practical application and knowledge of melanoma origin. Insolubility of these materials has led to diverse suggestions of pDA's structure. However, we found that pDA can be dissolved in certain ionic liquids, allowing us to identify its chemical structure as self-assembled supramolecular aggregates.
PHYSICAL CHEMISTRY CHEMICAL PHYSICS
(2023)