Publications
Found 79 publications
2025
AIMNet2: a neural network potential to meet your neutral, charged, organic, and elemental-organic needs
Chemical Science (2025)
The paper presents AIMNet2, an advanced machine learning interatomic potential designed to handle neutral, charged, organic, and elemental-organic molecules. Built upon previous work in neural network potentials, AIMNet2 is trained on extensive datasets to accurately predict energies and forces across a wide range of molecular systems. This development significantly enhances the ability to perform large-scale simulations with high accuracy, making it a valuable tool for computational chemistry research.
All That Glitters Is Not Gold: Importance of Rigorous Evaluation of Proteochemometric Models
Journal of Chemical Information and Modeling (2025)
The research evaluates proteochemometric models (PCMs) using kinase-ligand bioactivity prediction, focusing on factors like data curation, permutation testing, class imbalances, and data splitting strategies. Findings reveal that these elements significantly impact model performance, with protein embeddings contributing minimally as shown by permutation testing.
Anticipating the Selectivity of Intramolecular Cyclization Reaction Pathways with Neural Network Potentials
Journal of Chemical Theory and Computation (2025)
The authors present a method that integrates graph-based enumeration with machine learning techniques, specifically using the AIMNet2-rxn neural network potential, to identify relevant elementary steps in complex cyclization reactions. The approach efficiently filters intermediates and evaluates candidate pathways by estimating activation energies and predicting stereoselectivity. Validation demonstrates the model's ability to correctly anticipate reaction outcomes and replicate key enabling steps in natural product synthesis.
Proto-Yield: An Uncertainty-Aware Prototype Network for Yield Prediction in Real-world Chemical Reactions
(2025)
Proto-Yield addresses the challenge of predicting chemical reaction yields by incorporating uncertainty through a probabilistic approach. It models outcomes as distributions across high, medium, and low yield regimes, learning from noisy data to enhance prediction accuracy compared to existing regression-based methods.
Machine learning anomaly detection of automated HPLC experiments in the cloud laboratory
Digital Discovery (2025)
The study presents a transferable machine learning framework designed to detect anomalies in automated high-performance liquid chromatography (HPLC) experiments within a cloud laboratory environment. The system operates in real time, identifying affected runs and ensuring expert-level quality control without the need for human oversight. This approach enhances the reliability of autonomous experimentation by flagging unexpected events promptly, thereby maintaining data integrity and experimental accuracy.
Machine learning interatomic potentials at the centennial crossroads of quantum mechanics
Nature Computational Science (2025)
The paper explores the development of machine learning interatomic potentials through four main challenges: chemical accuracy, computational efficiency, interpretability, and universal generalizability. It highlights advancements in neural network architectures, physics-informed approaches that integrate domain knowledge into models, and foundation models trained on extensive datasets to achieve high accuracy while maintaining computational efficiency. These innovations aim to create predictive and transferable frameworks for next-generation molecular modeling.
Fast and Accurate Ring Strain Energy Predictions with Machine Learning and Application in Strain-Promoted Reactions
JACS Au (2025)
The study presents a machine learning-based approach utilizing AIMNet2 interatomic potentials and Auto3D for identifying low-energy conformers and predicting ring strain energy (RSE). The method achieves high accuracy with an R² of 0.997 and MAE of 0.896 kcal/mol compared to QM calculations, while being significantly faster. Applications include distinguishing reactive from nonreactive molecules in various reactions like click chemistry and polymerization, demonstrating broad applicability. Additionally, the creation of the RSE Atlas provides a comprehensive resource for studying strain effects across diverse molecular systems.
Efficient Molecular Crystal Structure Prediction and Stability Assessment with AIMNet2 Neural Network Potentials
Crystal Growth & Design (2025)
This study introduces a methodology that leverages target-specific AIMNet2 MLIPs to accelerate computational crystal structure prediction (CSP). By training these potentials on gas-phase dispersion-corrected DFT reference data of n-mers, the approach successfully extends to crystalline environments, accurately characterizing the CSP landscape and ranking structures by relative stability without requiring expensive periodic calculations.
GEOM-drugs revisited: toward more chemically accurate benchmarks for 3D molecule generation
Digital Discovery (2025)
The authors revisited the GEOM-drugs benchmark by correcting previous metrics and introducing a novel energy-based structural evaluation method. They developed new metrics to assess molecular conformations more accurately, leveraging machine learning potentials for energy calculations. Various generative models were tested against these benchmarks, providing insights into their performance and limitations in generating chemically accurate 3D structures.
ANI-1xBB: An ANI-Based Reactive Potential for Small Organic Molecules
Journal of Chemical Theory and Computation (2025)
ANI-1xBB is a novel reactive machine learning potential that enhances the prediction of reaction energetics and barrier heights by training on off-equilibrium molecular conformers. The model leverages an automated bond-breaking workflow to generate high-quality reactive data sets, improving transition state modeling and reaction pathway predictions. ANI-1xBB demonstrates strong performance across various reaction types, including pericyclic reactions and radical-driven processes.
Including Physics-Informed Atomization Constraints in Neural Networks for Reactive Chemistry
Journal of Chemical Information and Modeling (2025)
The authors present a technique to enhance MLIPs by incorporating physics-informed atomization constraints. This approach improves the prediction of energies in systems with isolated atoms, crucial for reactive processes. The developed models, such as HIP-NN-AE and ANI-AE, demonstrate consistent performance without negatively impacting other tasks.
High-throughput electronic property prediction of cyclic molecules with 3D-enhanced machine learning
Chemical Science (2025)
The research utilizes AIMNet2 with 3D molecular representations to predict electronic properties of cyclic molecules from the Ring Vault dataset. The model's performance was compared against 2D approaches, demonstrating superior accuracy in predicting electronic properties across a diverse set of cyclic compounds. This highlights the importance of incorporating three-dimensional structural information for accurate property prediction in complex molecular systems.
2024
In silico screening of LRRK2 WDR domain inhibitors using deep docking and free energy simulations
Chemical Science (2024)
The study employed Deep Docking to identify potential inhibitors targeting the WDR domain of LRRK2, followed by free energy molecular dynamics (MD) simulations to evaluate their binding affinities. Experimental validation confirmed the effectiveness of selected candidates, demonstrating the utility of this computational approach in drug discovery.
ANI/EFP: Modeling Long-Range Interactions in ANI Neural Network with Effective Fragment Potentials
Journal of Chemical Theory and Computation (2024)
The authors propose integrating atomic electrostatic potentials generated via polarizable effective fragment potentials (EFP) into the ANI neural network framework. This approach involves retraining the ANI model with additional input features derived from EFP, enabling more accurate predictions of solute-solvent interaction energies. The method demonstrates high accuracy on trained datasets and shows promise for extrapolating to untrained solvent environments, suggesting potential for broader applications in molecular modeling.
Discovery of Crystallizable Organic Semiconductors with Machine Learning
J. Am. Chem. Soc. (2024)
The study employed machine learning algorithms to estimate the melting point (Tm) and crystallization driving force (ΔGc) of organic molecules, aiming to identify those capable of forming platelet morphologies. Six candidate molecules were experimentally evaluated, resulting in three that formed platelets, one that formed spherulites, and two that resisted crystallization. This demonstrates the effectiveness of ML in predicting thermal properties and highlights Tm and ΔGc as critical metrics for crystallization behavior prediction.
De novo molecule design towards biased properties via a deep generative framework and iterative transfer learning
Digital Discovery (2024)
The authors developed the RRCGAN model using an iterative transfer learning approach to generate chemically valid molecules with targeted energy gap values. Validation through density functional theory (DFT) demonstrated that 75% of the generated molecules achieved a relative error (RE) of less than 20% compared to the target energy gaps, indicating successful property biasing in molecular design.
Integrating QSAR modelling and deep learning in drug discovery: the emergence of deep QSAR
Nat. Rev. Drug Discov. (2024)
The publication discusses the integration of QSAR modeling with deep learning techniques, leading to the development of deep QSAR methods. These methods include deep generative models and reinforcement learning for molecular design, as well as applications in synthetic planning and virtual screening. The authors also explore the potential impact of quantum computing on accelerating deep QSAR applications and emphasize the importance of open-source resources for advancing computer-aided drug design.
Exploring the frontiers of condensed-phase chemistry with a general reactive machine learning potential
Nature Chemistry (2024)
The study presents ANI-1xnr, a general reactive machine learning interatomic potential developed using automated sampling of condensed-phase reactions. This MLIP was validated across five distinct systems, including carbon solid-phase nucleation, graphene ring formation, biofuel additives, methane combustion, and glycine formation from early earth molecules. The results demonstrate that ANI-1xnr accurately predicts reaction outcomes, matching experimental data and traditional model chemistry methods. This advancement enables efficient high-throughput in silico experimentation for reactive chemistry involving C, H, N, and O elements.
2023
Generative Models as an Emerging Paradigm in the Chemical Sciences
J. Am. Chem. Soc. (2023)
The authors discuss how generative models, including GANs, VAEs, flow, and diffusion models, can address the challenges of traditional computational approaches by enabling the generation of novel compounds with desired properties. They highlight the key differences between these models, recent success stories in chemical applications, and remaining challenges for practical implementation.
Machine Learning Interatomic Potentials and Long-Range Physics
J. Phys. Chem. A (2023)
The authors discuss methods to improve MLIPs by addressing long-range interactions, including dispersion corrections, charge prediction using atomic descriptors, self-consistency iterations, message passing for system information propagation, and equilibration schemes. These approaches aim to overcome the limitations of short-range models in accurately describing complex systems.
Themed collection on Insightful Machine Learning for Physical Chemistry
Physical Chemistry Chemical Physics (2023)
The collection presents a variety of methodologies and results from different studies, highlighting the application of machine learning in understanding complex chemical systems. It includes insights into neural network potentials, feature engineering for chemical data, and optimization strategies for machine learning models in quantum chemistry contexts.
Active Learning Guided Drug Design Lead Optimization Based on Relative Binding Free Energy Modeling
J. Chem. Inf. Model. (2023)
The study presents an AL-AutoML workflow that efficiently identifies compounds with low BFE among congeneric ligands. Applied to SARS-CoV-2 protease inhibitors, it found 133 compounds with improved affinity, including 16 with over a 100-fold improvement, outperforming traditional methods.
Structure Prediction of Epitaxial Organic Interfaces with Ogre, Demonstrated for Tetracyanoquinodimethane (TCNQ) on Tetrathiafulvalene (TTF)
J. Phys. Chem. C (2023)
The authors present Ogre, a computational framework leveraging machine learning potentials to predict the structure of epitaxial organic interfaces. By applying Ogre to the TCNQ-on-TTF system, they successfully modeled the interface structure and validated it experimentally. This approach provides insights into the charge transfer properties at organic heterojunctions, which are critical for designing new materials in organic electronics.
2022
Learning molecular potentials with neural networks
WIREs Comput. Mol. Sci. (2022)
The paper provides a comprehensive overview of neural network potentials, discussing their accuracy compared to quantum mechanical methods, strategies for improvement, and applications in various fields including drug discovery. It highlights how these models can bridge the gap between computational cost and reliability, making them valuable tools for large-scale studies.
Simulations of Pathogenic E1α Variants: Allostery and Impact on Pyruvate Dehydrogenase Complex-E1 Structure and Function
Journal of Chemical Information and Modeling (2022)
Using molecular dynamics simulations, the study examines how specific E1α mutations affect the structure and function of PDC-E1. It identifies changes in phosphorylation loops and communication networks within the protein, suggesting allosteric effects that may influence enzyme activity and provide targets for therapeutic intervention.
Prediction of protein pKawith representation learning
Chemical Science (2022)
The study presents a novel empirical machine learning approach for predicting protein pKa values, leveraging representation learning techniques. The model achieves mean absolute errors (MAEs) below 0.5 for all amino acid types, demonstrating significant improvement over existing methods. This advancement provides a reliable tool for understanding protein stability and function in various biochemical contexts.
Auto3D: Automatic Generation of the Low-Energy 3D Structures with ANI Neural Network Potentials
J. Chem. Inf. Model. (2022)
The Auto3D package automates the generation of low-energy 3D molecular structures by handling stereoisomers and conformers. It uses SMILES as input, performs isomer enumeration, duplicate filtering, 3D building, geometry optimization, and ranking. The ANI-2xt model extension improves accuracy for tautomeric reactions, reducing energy calculation errors compared to previous models.
The transformational role of GPU computing and deep learning in drug discovery
Nature Machine Intelligence (2022)
The study investigates the role of GPU computing and deep learning in enhancing drug discovery, specifically focusing on Alzheimer's Disease. By analyzing CSF biomarkers from the ADNI database using statistical methods and machine learning, the authors identify discordant cases between CSF Aβ42:40 levels and amyloid status, highlighting the need for detailed evaluation before clinical trials.
2021
Best practices in machine learning for chemistry
Nature Chemistry (2021)
This publication discusses the essential guidelines and methodologies for effectively integrating machine learning into chemical research. It likely explores different ML techniques, their applications in areas such as potential energy surfaces and quantum calculations, and offers practical advice on implementation and validation strategies to ensure robust and reliable results.
Harnessing the Power of Smart and Connected Health to Tackle COVID-19: IoT, AI, Robotics, and Blockchain for a Better World
IEEE Internet of Things Journal (2021)
This article discusses how advanced technologies such as IoT, AI, robotics, and blockchain have been leveraged to combat the COVID-19 pandemic. It highlights applications in contact tracing, remote patient monitoring, vaccine development, and data integrity, providing a comprehensive overview of technological contributions to public health management during crises.
OpenChem: A Deep Learning Toolkit for Computational Chemistry and Drug Design
Journal of Chemical Information and Modeling (2021)
The authors developed OpenChem, a deep learning toolkit built on PyTorch, designed to facilitate model development in computational chemistry and drug design. The toolkit features a modular architecture, allowing users to easily integrate various components, and includes several data preprocessing modules that streamline the handling of chemical datasets. By providing these tools, OpenChem aims to lower the barrier for researchers to implement deep learning models in their work.
A critical overview of computational approaches employed for COVID-19 drug discovery
Chemical Society Reviews (2021)
This publication evaluates various computational approaches employed in the drug discovery process for COVID-19, including molecular docking, QSAR modeling, and machine learning techniques. It discusses case studies that illustrate successful applications of these methods and addresses challenges such as data quality, model validation, and translatability to clinical settings.
Machine-Learning-Guided Discovery of 19 F MRI Agents Enabled by Automated Copolymer Synthesis
Journal of the American Chemical Society (2021)
The study developed an automated flow synthesis platform to iteratively synthesize 397 unique copolymer compositions within a six-variable space. Machine learning identified nonintuitive design criteria, enabling the discovery of over ten copolymers that outperformed existing MRI agents by exploring less than 0.9% of the compositional space.
Development of Multimodal Machine Learning Potentials: Toward a Physics-Aware Artificial Intelligence
Accounts of Chemical Research (2021)
The authors introduce ANI and AIMNet, neural network potentials that leverage large datasets and symmetry functions. They extend these models with transfer learning and active learning to improve data selection. Additionally, they present ML-EHM, which combines ML with the extended Hückel method, demonstrating improved accuracy in describing molecular orbital behavior through bond rotations.
2020
Extending the Applicability of the ANI Deep Learning Molecular Potential to Sulfur and Halogens
Journal of Chemical Theory and Computation (2020)
The authors extended the ANI-1x model to create ANI-2x, incorporating sulfur, fluorine, and chlorine. This new model was refined for torsional predictions and tested across various benchmarks, demonstrating high accuracy and efficiency in predicting molecular energies compared to DFT, making it valuable for drug development applications.
TorchANI: A Free and Open Source PyTorch-Based Deep Learning Implementation of the ANI Neural Network Potentials
Journal of Chemical Information and Modeling (2020)
TorchANI provides a user-friendly, cross-platform alternative to NeuroChem for training and inference of ANI models. By leveraging PyTorch's autograd engine, it enables automatic computation of forces and Hessians, as well as force training without additional code. This implementation sacrifices some performance for ease of use and flexibility.
Correction: QSAR without borders
Chemical Society Reviews (2020)
The authors present advanced QSAR models that integrate diverse data sources and employ machine learning techniques to predict molecular interactions more effectively. The study validates these models against existing datasets, demonstrating improved predictive capabilities compared to traditional methods.
DRACON: disconnected graph neural network for atom mapping in chemical reactions
Physical Chemistry Chemical Physics (2020)
The study introduces DRACON, a novel approach that applies disconnected graph neural networks to solve the atom mapping problem in chemical reactions. By formulating the task as node-classification within a disconnected graph framework, the authors generalize traditional graph convolutional neural networks (GCNNs) to handle such structures effectively. This method enables accurate mapping of atoms across source and product molecules, which is crucial for predicting reaction outcomes and understanding mechanistic pathways.
2019
Inter-Modular Linkers play a crucial role in governing the biosynthesis of non-ribosomal peptides
Bioinformatics (2019)
The study explores how IMLs facilitate communication between NRPS modules during peptide synthesis. By developing a parser to extract IMLs from bacterial genomes, the authors reveal that these linkers exhibit substrate specificity, crucial for successful module interactions in synthetic peptide production.
Quantitative Structure–Price Relationship (QS$R) Modeling and the Development of Economically Feasible Drug Discovery Projects
Journal of Chemical Information and Modeling (2019)
The study presents the development of a quantitative structure-price relationship (QS$R) model using machine learning. This model was trained on a diverse dataset of chemicals, ranging in price from $20 to over $10,000, enabling predictions of molecule costs based on their structural features. The QS$R model integrates cost considerations into the drug discovery pipeline, aiding in the selection of economically viable compounds early in the process.
Text mining facilitates materials discovery
Nature (2019)
The study discusses the application of text mining to analyze vast amounts of scientific literature, identifying patterns and relationships that could lead to the discovery of new materials. By leveraging natural language processing and machine learning algorithms, the authors propose a method to extract meaningful information from textual data, which can guide experimental efforts and accelerate the development of novel materials with desired properties.
Adsorption of nitrogen-containing compounds on hydroxylated α-quartz surfaces
RSC Advances (2019)
The study employs computational chemistry techniques to determine the adsorption energies of various nitrogen-containing compounds on two distinct hydroxylated α-quartz surfaces. By analyzing these interactions, the research identifies key factors influencing molecular adsorption, such as surface structure and chemical composition. The findings provide insights into how different functional groups interact with quartz surfaces, which is crucial for understanding environmental and industrial processes involving these materials.
2018
Machine learning for molecular and materials science
Nature (2018)
The authors provide an overview of machine learning applications in molecular and materials science, discussing various techniques suitable for addressing research questions in this domain. They highlight the potential for accelerating design, synthesis, characterization, and application processes through artificial intelligence, touching on areas such as neural network potentials, force fields, and crystal structure prediction.
Materials discovery by chemical analogy: role of oxidation states in structure prediction
Faraday Discussions (2018)
The study presents a computational framework that uses chemical analogy to predict the likelihood of hypothetical compounds forming, focusing on the role of oxidation states. The model evaluates the probability of compound stability by analyzing the oxidation states of constituent elements and their compatibility in known structures. This approach was validated against existing materials data, demonstrating its effectiveness in identifying plausible new compounds.
Diffusion of energetic compounds through biological membrane: Application of classical MD and COSMOmic approximations
Journal of Biomolecular Structure and Dynamics (2018)
The research employs MD simulations and the COSMOmic method to evaluate how energetic compounds like TNT, DNT, DNAN, and NTO diffuse through a POPC lipid bilayer. Factors such as ionic strength and DMSO presence were considered. MD provided detailed free energy profiles, while COSMOmic was effective for predicting log(K_lipw) values. The findings highlight the influence of functional groups on compound distribution within the membrane.
AFLOW-ML: A RESTful API for machine-learning predictions of materials properties
Computational Materials Science (2018)
The paper introduces AFLOW-ML, a RESTful API designed to facilitate machine learning predictions of various materials properties. By leveraging pre-trained models from the AFLOWLIB database, this tool allows users to query properties such as elastic constants, band gaps, and other material characteristics through a simple web interface. The implementation emphasizes ease of use and integration with existing workflows, making advanced computational materials science more accessible to researchers across disciplines.
Transferable Dynamic Molecular Charge Assignment Using Deep Neural Networks
Journal of Chemical Theory and Computation (2018)
This research presents a machine learning approach using the HIP-NN model to predict dynamic molecular charges efficiently. The method demonstrates high accuracy across various molecules and charge partitioning schemes, outperforming traditional DFT calculations in speed while maintaining precision.
Discovering a Transferable Charge Assignment Model Using Machine Learning
The Journal of Physical Chemistry Letters (2018)
The study introduces the Affordable Charge Assignment (ACA) model, which uses machine learning to predict partial atomic charges by replicating molecular dipole moments across diverse conformations. The ACA model is computationally inexpensive and demonstrates high accuracy in predicting dipoles for out-of-sample molecules. It also shows transferability to larger molecules, enabling accurate predictions of dipole and quadrupole moments beyond the training set. Applications include dynamical trajectories of biomolecules and infrared spectrum analysis.
Transforming Computational Drug Discovery with Machine Learning and AI
ACS Medicinal Chemistry Letters (2018)
The paper explores the current progress of applying machine learning and artificial intelligence to address challenges in computational drug discovery. It identifies key areas where these technologies can enhance pharmaceutical research, potentially leading to more efficient drug development processes.
2017
ANI-1: an extensible neural network potential with DFT accuracy at force field computational cost
Chemical Science (2017)
The authors present ANI-1, a deep neural network potential trained on quantum mechanical energies calculated using density functional theory (DFT). This potential is designed to be both accurate and computationally efficient, enabling the study of organic molecules containing H, C, N, and O atoms. The methodology involves training the neural network on a dataset of DFT energies, resulting in a model that can predict atomic interactions with high accuracy while maintaining the computational speed typical of classical force fields.
2016
Atlas Regeneration Company, Inc.
Regenerative Medicine (2016)
The study presents the Universal Signalome Atlas, a comprehensive map of stem cell differentiation pathways. This atlas integrates multiomics data to analyze pathway activation states across various stem cells and their differentiated products. Applications include quality assurance for engineered cell products and directed regeneration pharmacology, where compounds are screened to efficiently convert pluripotent cells into desired subtypes. Additionally, the Regeneration Intelligence system is developed to predict and determine stem cell signaling pathways, aiding in specific differentiation directions.
Material informatics driven design and experimental validation of lead titanate as an aqueous solar photocathode
Materials Discovery (2016)
The researchers employed computational methods within the framework of materials informatics to identify lead titanate as a potential candidate for use in solar water splitting applications. This was followed by experimental validation, confirming its effectiveness as a photocathode under aqueous conditions. The approach highlights the synergy between computational design and experimental verification in accelerating materials discovery.
2015
Materials Cartography: Representing and Mining Materials Space Using Structural and Electronic Fingerprints
Chemistry of Materials (2015)
The study introduces novel analytical approaches using structural and electronic fingerprints to represent and mine materials databases. These fingerprints enable querying large datasets, mapping the connectivity of materials space (materials cartography), and developing predictive models for material properties. The framework is demonstrated by modeling critical temperatures of superconductors, showcasing its utility in materials discovery and design.
Are the reduction and oxidation properties of nitrocompounds dissolved in water different from those produced when adsorbed on a silica surface? A DFT M05-2X computational study
Journal of Computational Chemistry (2015)
The research employs DFT M05-2X computational methods with cluster models to analyze how adsorption on silica and solvation in water affect the redox properties of nitrocompounds. The study uses PCM(Pauling) and SMD solvation models to evaluate changes in electron affinity, ionization potential, and Gibbs free energy, revealing that solvation has a more significant impact than adsorption on promoting redox transformations.
2012
Validation of a novel secretion modification region (SMR) of HIV-1 Nef using cohort sequence analysis and molecular modeling
Journal of Molecular Modeling (2012)
The research identifies the SMR motif in Nef through computational methods, demonstrating its structural importance for secretion. Sequence analysis across patient cohorts confirmed high conservation of this region, supporting its functional relevance.
Mechanical properties of silicon nanowires
WIREs Computational Molecular Science (2012)
The paper explores the mechanical properties of silicon nanowires (SiNWs) through both experimental and theoretical approaches. It highlights how improved potentials and advanced modeling techniques, including quantum-mechanical effects and temperature considerations, clarify the size-dependent behavior of Young's modulus and ductile-brittle transitions in SiNWs. The authors also discuss postprocessing techniques to mitigate errors in experimental results.
In silico structure–function analysis of E. cloacae nitroreductase
Proteins: Structure, Function, and Bioinformatics (2012)
The researchers performed molecular dynamics simulations of EcNR in three states: oxidized, reduced with benzoate inhibitor, and reduced with nitrobenzene. Principal Component Analysis revealed increased flexibility in the enzyme upon complexation, particularly in helix H6 near the binding site. A multiple sequence alignment identified conserved regions within the FMN binding site, leading to a proposed new catalytic mechanism.
2011
Evaluation of natural and nitramine binding energies to 3-D models of the S1S2 domains in the N-methyl-D-aspartate receptor
Journal of Molecular Modeling (2011)
The study uses docking methods and REMD simulations to assess how natural and nitramine compounds bind to the NMDA receptor's S1S2 domains, providing insights into potential drugs or toxicants.
Car–Parrinello Molecular Dynamics Simulations of Tensile Tests on Si⟨001⟩ Nanowires
The Journal of Physical Chemistry C (2011)
The study employs Car–Parrinello molecular dynamics to simulate tensile tests on Si⟨001⟩ nanowires, investigating the effects of H-passivation and wire geometry on mechanical behavior. The simulations reveal that surface layer composition, particularly the presence of SiH2 groups, and wire shape significantly influence Young's modulus and strain tolerance. Comparisons between octahedral and tetrahedral wire shapes highlight the role of {100} facets in strain relaxation during tension. Structural parameter changes provide insights into atomic motion patterns leading to fracture.
Novel view on the mechanism of water-assisted proton transfer in the DNA bases: bulk water hydration
Physical Chemistry Chemical Physics (2011)
The paper uses classical and Car-Parrinello molecular dynamics simulations to investigate the stability of nucleic acid base-water bridge complexes. They find that these complexes exist for significantly shorter durations than predicted by static ab initio methods, which impacts the feasibility of proton transfer. Based on this dynamic analysis, new rate constants for tautomerization are proposed and applied to assess water-assisted mechanisms in biological systems such as E. coli cells.
Effect of Solvation on the Vertical Ionization Energy of Thymine: From Microhydration to Bulk
The Journal of Physical Chemistry A (2011)
This study examines how hydration affects thymine's VIE by comparing microsolvated clusters with bulk water conditions. Using EOM-IP-CCSD and EFP methods, the research highlights that microsolvation reduces IE while bulk solvation increases it, offering a comprehensive understanding of solvation effects on molecular properties.
Toward robust computational electrochemical predicting the environmental fate of organic pollutants
Journal of Computational Chemistry (2011)
The study employs density functional theory (DFT) with various functionals and solvation models to calculate electron attachment free energies and solvation effects. The methodology accurately predicts reduction potentials, validated against experimental data, and applies these findings to evaluate the reactivity of iron species in reducing organic pollutants under different pH conditions.
2010
Hydration of nucleic acid bases: a Car–Parrinello molecular dynamics approach
Physical Chemistry Chemical Physics (2010)
This research employs Car-Parrinello molecular dynamics (CPMD) to comprehensively analyze the hydration interactions of nucleic acid bases in bulk water. The study details aspects such as coordination numbers, hydrogen bonding patterns, and the lifetimes of water molecules in the first hydration shell, offering a deeper understanding of how these biological molecules interact with aqueous environments.
New insight on structural properties of hydrated nucleic acid bases from ab initio molecular dynamics
Physical Chemistry Chemical Physics (2010)
The researchers performed Car-Parrinello Molecular Dynamics (CPMD) simulations of nucleic acid bases in bulk water to investigate how hydration affects their structure and flexibility. The results showed that hydration leads to bond length alterations, zwitter-ionic resonance structures, restricted amino group mobility, and increased planar-like conformations. Additionally, the dynamic aqueous environment caused all nucleic acid base rings to become equally flexible, resulting in a non-planar effective conformation despite planar geometry being energetically favorable.
Reaction of bicyclo[2.2.1]hept‐5‐ene‐endo‐2‐ylmethylamine and nitrophenyl glycidyl ethers
Journal of Physical Organic Chemistry (2010)
The research combines experimental techniques, including chromatography and spectroscopy, with quantum-chemical calculations at the B3LYP/6-311+G(d,p) level. The authors analyze the regioselective aminolysis of epoxides and the minor products resulting from aryl nucleophilic substitution. Quantum calculations are used to evaluate activation barriers and free Gibbs energies, confirming the abnormal reaction course.
One-electron standard reduction potentials of nitroaromatic and cyclic nitramine explosives
Environmental Pollution (2010)
The research employs linear free energy relationships experimentally and ab initio calculations theoretically to determine the standard reduction potentials (Eo(R-NO2/R-NO2-)) of various explosive compounds. The results show a decreasing trend in Eo values from di- and tri-nitroaromatics to nitramines and nitroimino compounds, correlating with their reductive degradation rates under anoxic conditions. This thermodynamic control provides insights into the environmental fate of these explosives.
2008
Ab Initio Molecular Dynamics Study on the Initial Chemical Events in Nitramines: Thermal Decomposition of CL-20
The Journal of Physical Chemistry B (2008)
Using ab initio molecular dynamics simulations, the paper examines the initial stages of CL-20's thermal decomposition. It identifies homolysis of the N-NO2 bond as the primary reaction channel and provides insights into product populations and activation barriers, contributing to understanding explosive behavior.
Efficient and accurate ab initio prediction of thermodynamic parameters for intermolecular complexes
Chemical Physics Letters (2008)
The paper presents a computational approach to predict thermodynamic parameters such as enthalpy and entropy changes for intermolecular complexes using ab initio quantum chemical calculations. The methodology involves calculating electronic structure properties at various levels of theory, which are then used to derive thermodynamic quantities. The results demonstrate high accuracy compared to experimental data, making it a reliable tool for studying molecular interactions.
2007
Are Isolated Nucleic Acid Bases Really Planar? A Car−Parrinello Molecular Dynamics Study
The Journal of Physical Chemistry B (2007)
The study employs Car-Parrinello molecular dynamics to investigate the conformational flexibility of isolated DNA bases. By comparing computational methods, it establishes that DFT with plane wave basis sets accurately models pyrimidine ring deformability. Simulations reveal significant nonplanar conformations, challenging the assumption of planarity in nucleic acid bases.
Theoretical calculations: Can Gibbs free energy for intermolecular complexes be predicted efficiently and accurately?
Journal of Computational Chemistry (2007)
The study assesses various quantum-chemical methods (MP2, CCSD(T), DFT) to predict Gibbs free energy accurately. It finds that extrapolating MP2 and CCSD(T) energies to complete basis sets is essential for high accuracy. Additionally, it explores CPMD simulations and thermodynamic integration techniques, highlighting their capabilities in achieving accurate results within 1-3 kcal/mol.
Carboxamides and amines having two and three adamantane fragments
Russian Journal of Organic Chemistry (2007)
The authors synthesized novel organic compounds featuring multiple adamantane groups, which are known for their rigid and stable structures. The study focuses on the chemical synthesis methods used to create these compounds, as well as their characterization through various analytical techniques. The research contributes to the understanding of how adamantane fragments can be incorporated into larger molecular frameworks, potentially leading to new materials with unique properties.
2006
Structure-toxicity relationships of nitroaromatic compounds: Full-length paper
Molecular Diversity (2006)
The study investigates the relationship between the structure and toxicity of nitroaromatic compounds using QSAR modeling. Computational methods were employed to analyze molecular descriptors, which were then used to build predictive models for toxicity. The findings provide insights into how specific structural features influence toxicological outcomes, contributing to a better understanding of environmental impacts and potential applications in drug design.
2005
Acylation of Aminopyridines and Related Compounds with Endic Anhydride
Russian Journal of Organic Chemistry (2005)
The paper explores the acylation process of aminopyridines and related compounds with Endic anhydride. The authors likely examined various reaction conditions, studied the mechanism in detail, and identified key factors influencing the acylation efficiency. Their findings contribute to understanding how these reactions proceed under different experimental setups.
Synthesis and Reactivity of Amines Containing Several Cage-like Fragments
Russian Journal of Organic Chemistry (2005)
This research investigates the synthesis methods and subsequent reactivity of amines that incorporate multiple cage-like fragments. The work likely delves into the structural properties of these compounds and examines how their unique architectures influence chemical reactions. Without an abstract, specific details on the methodologies or findings are unavailable, but the focus appears to be on advancing knowledge in organic chemistry through the exploration of novel amine structures.
2004
Amides containing two norbornene fragments. Synthesis and chemical transformations
Russian Journal of Organic Chemistry (2004)
This research investigates the synthesis and subsequent chemical transformations of amides that incorporate two norbornene fragments. The authors detail various synthetic routes and examine the resulting compounds' properties, contributing to the understanding of these specific organic structures and their potential applications.
Reaction of Endic Anhydride with Hydrazines and Acylhydrazines
Russian Journal of Organic Chemistry (2004)
This study details the cycloaddition reactions between imine derivatives and acyl chlorides, leading to the formation of cis-2-azetidinone stereoisomers. The compounds were characterized using spectroscopic methods, contributing to the understanding of β-lactam synthesis pathways.
Modeling the Gas-Phase Reduction of Nitrobenzene to Nitrosobenzene by Iron Monoxide: A Density Functional Theory Study
The Journal of Physical Chemistry A (2004)
The authors employ density functional theory (DFT) to investigate the reaction mechanism of nitrobenzene reduction to nitrosobenzene catalyzed by iron monoxide. They calculate the electronic structure of reactants, transition states, and products to determine the most plausible reaction pathway. The study identifies key intermediates and transition states, providing insights into the role of iron monoxide in facilitating this transformation.
2003
Amino Alcohols with Bicyclic Carbon Skeleton. Alternative Functionalization of Nucleophilic Reaction Centers
Russian Journal of Organic Chemistry (2003)
This study presents a new approach to synthesizing cis-2-azetidinones by reacting imines with acyl chlorides in the presence of triphenylamine. The authors characterize the products using spectroscopic methods and discuss their potential applications in organic chemistry.
2002
New N-(Arylsulfonyl)-5-aminomethylbicyclo[2.2.1]hept-2-enes. Synthesis, 1H and 13C NMR Spectra, and Chemical Reactions
Russian Journal of Organic Chemistry (2002)
The authors synthesized a series of N-(Arylsulfonyl)-5-aminomethylbicyclo[2.2.1]hept-2-enes and characterized them using ¹H and ¹³C NMR spectroscopy. The chemical reactivity of these compounds was also investigated, providing insights into their structural properties and potential applications in organic chemistry.