Publications

In this work, we combined Deep Docking and free energy MD simulations for the in silico screening and experimental validation for potential inhibitors of leucine rich repeat kinase 2 (LRRK2) targeting the WD40 repeat (WDR) domain.

The RRCGAN, validated through DFT, demonstrates success in generating chemically valid molecules targeting energy gap values with 75% of the generated molecules have RE of <20% of the targeted values.

Abstract Atomistic simulation has a broad range of applications from drug design to materials discovery. Machine learning interatomic potentials (MLIPs) have become an efficient alternative to computationally expensive ab initio simulations. For this reason, chemistry and materials science would greatly benefit from a general reactive MLIP, that is, an MLIP that is applicable to a broad range of reactive chemistry without the need for refitting. Here we develop a general reactive MLIP (ANI-1xnr) through automated sampling of condensed-phase reactions. ANI-1xnr is then applied to study five distinct systems: carbon solid-phase nucleation, graphene ring formation from acetylene, biofuel additives, combustion of methane and the spontaneous formation of glycine from early earth small molecules. In all studies, ANI-1xnr closely matches experiment (when available) and/or previous studies using traditional model chemistry methods. As such, ANI-1xnr proves to be a highly general reactive MLIP for C, H, N and O elements in the condensed phase, enabling high-throughput in silico reactive chemistry experimentation.

This themed collection includes a collection of articles on Insightful Machine Learning for Physical Chemistry.

Catalyzed by enormous success in the industrial sector, many research programs have been exploring data-driven, machine learning approaches. Performance can be poor when the model is extrapolated to new regions of chemical space, e.g., new bonding types, new many-body interactions. Another important limitation is the spatial locality assumption in model architecture, and this limitation cannot be overcome with larger or more diverse datasets. The outlined challenges are primarily associated with the lack of electronic structure information in surrogate models such as interatomic potentials. Given the fast development of machine learning and computational chemistry methods, we expect some limitations of surrogate models to be addressed in the near future; nevertheless spatial locality assumption will likely remain a limiting factor for their transferability. Here, we suggest focusing on an equally important effort—design of physics-informed models that leverage the domain knowledge and employ machine learning only as a corrective tool. In the context of material science, we will focus on semi-empirical quantum mechanics, using machine learning to predict corrections to the reduced-order Hamiltonian model parameters. The resulting models are broadly applicable, retain the speed of semiempirical chemistry, and frequently achieve accuracy on par with much more expensive ab initio calculations. These early results indicate that future work, in which machine learning and quantum chemistry methods are developed jointly, may provide the best of all worlds for chemistry applications that demand both high accuracy and high numerical efficiency.

Deep-HP is a scalable extension of the Tinker-HP multi-GPU molecular dynamics (MD) package enabling the use of Pytorch/TensorFlow Deep Neural Network (DNN) models.

A sensitive model captures the reactivity cliffs but overfit to yield outliers. On the other hand, a robust model disregards the yield outliers but underfits the reactivity cliffs.

Newly developed Δ 2 -learning models enable state-of-the-art accuracy in predicting the properties of chemical reactions.

AbstractThe potential energy of molecular species and their conformers can be computed with a wide range of computational chemistry methods, from molecular mechanics to ab initio quantum chemistry. However, the proper choice of the computational approach based on computational cost and reliability of calculated energies is a dilemma, especially for large molecules. This dilemma is proved to be even more problematic for studies that require hundreds and thousands of calculations, such as drug discovery. On the other hand, driven by their pattern recognition capabilities, neural networks started to gain popularity in the computational chemistry community. During the last decade, many neural network potentials have been developed to predict a variety of chemical information of different systems. Neural network potentials are proved to predict chemical properties with accuracy comparable to quantum mechanical approaches but with the cost approaching molecular mechanics calculations. As a result, the development of more reliable, transferable, and extensible neural network potentials became an attractive field of study for researchers. In this review, we outlined an overview of the status of current neural network potentials and strategies to improve their accuracy. We provide recent examples of studies that prove the applicability of these potentials. We also discuss the capabilities and shortcomings of the current models and the challenges and future aspects of their development and applications. It is expected that this review would provide guidance for the development of neural network potentials and the exploitation of their applicability.This article is categorized under:Data Science > Artificial Intelligence/Machine LearningMolecular and Statistical Mechanics > Molecular InteractionsSoftware > Molecular Modeling

The behavior of proteins is closely related to the protonation states of the residues. Therefore, prediction and measurement of pKa are essential to understand the basic functions of proteins. In this work, we develop a new empirical scheme for protein pKa prediction that is based on deep representation learning. It combines machine learning with atomic environment vector (AEV) and learned quantum mechanical representation from ANI-2x neural network potential (J. Chem. Theory Comput. 2020, 16, 4192). The scheme requires only the coordinate information of a protein as the input and separately estimates the pKa for all five titratable amino acid types. The accuracy of the approach was analyzed with both cross-validation and an external test set of proteins. Obtained results were compared with the widely used empirical approach PROPKA. The new empirical model provides accuracy with MAEs below 0.5 for all amino acid types. It surpasses the accuracy of PROPKA and performs significantly better than the null model. Our model is also sensitive to the local conformational changes and molecular interactions.

The behavior of proteins is closely related to the protonation states of the residues. Therefore, prediction and measurement of pKa are essential to understand the basic functions of proteins. In this work, we develop a new empirical scheme for protein pKa prediction that is based on deep representation learning. It combines machine learning with atomic environment vector (AEV) and learned quantum mechanical representation from ANI-2x neural network potential (J. Chem. Theory Comput. 2020, 16, 4192). The scheme requires only the coordinate information of a protein as the input and separately estimates the pKa for all five titratable amino acid types. The accuracy of the approach was analyzed with both cross-validation and an external test set of proteins. Obtained results were compared with the widely used empirical approach PROPKA. The new empirical model provides accuracy with MAEs below 0.5 for all amino acid types. It surpasses the accuracy of PROPKA and performs significantly better than the null model. Our model is also sensitive to the local conformational changes and molecular interactions.

We developed new empirical ML model for protein pKaprediction with MAEs below 0.5 for all amino acid types.

AbstractDeep generative neural networks have been used increasingly in computational chemistry for de novo design of molecules with desired properties. Many deep learning approaches employ reinforcement learning for optimizing the target properties of the generated molecules. However, the success of this approach is often hampered by the problem of sparse rewards as the majority of the generated molecules are expectedly predicted as inactives. We propose several technical innovations to address this problem and improve the balance between exploration and exploitation modes in reinforcement learning. In a proof-of-concept study, we demonstrate the application of the deep generative recurrent neural network architecture enhanced by several proposed technical tricks to design inhibitors of the epidermal growth factor (EGFR) and further experimentally validate their potency. The proposed technical solutions are expected to substantially improve the success rate of finding novel bioactive compounds for specific biological targets using generative and reinforcement learning approaches.

Abstract Despite decades of intensive search for compounds that modulate the activity of particular protein targets, a large proportion of the human kinome remains as yet undrugged. Effective approaches are therefore required to map the massive space of unexplored compound–kinase interactions for novel and potent activities. Here, we carry out a crowdsourced benchmarking of predictive algorithms for kinase inhibitor potencies across multiple kinase families tested on unpublished bioactivity data. We find the top-performing predictions are based on various models, including kernel learning, gradient boosting and deep learning, and their ensemble leads to a predictive accuracy exceeding that of single-dose kinase activity assays. We design experiments based on the model predictions and identify unexpected activities even for under-studied kinases, thereby accelerating experimental mapping efforts. The open-source prediction algorithms together with the bioactivities between 95 compounds and 295 kinases provide a resource for benchmarking prediction algorithms and for extending the druggable kinome.

The bandgap is one of the most fundamental properties of condensed matter. However, an accurate calculation of its value, which could potentially allow experimentalists to identify materials suitable for device applications, is very computationally expensive. Here, active machine learning algorithms are used to leverage a limited number of accurate density functional theory calculations to robustly predict the bandgap of a very large number of novel 2D heterostructures. Using this approach, a database of ≈2.2 million bandgap values for various novel 2D van der Waals heterostructures is produced.

The behavior of proteins is closely related to the protonation states of the residues. Therefore, prediction and measurement of pKa are essential to understand the basic functions of proteins. In this work, we develop a new empirical scheme for protein pKa prediction that is based on deep representation learning. It combines machine learning with atomic environment vector (AEV) and learned quantum mechanical representation from ANI-2x neural network potential (J. Chem. Theory Comput. 2020, 16, 4192). The scheme requires only the coordinate information of a protein as the input and separately estimates the pKa for all five titratable amino acid types. The accuracy of the approach was analyzed with both cross-validation and an external test set of proteins. Obtained results were compared with the widely used empirical approach PROPKA. The new empirical model provides accuracy with MAEs below 0.5 for all amino acid types. It surpasses the accuracy of PROPKA and performs significantly better than the null model. Our model is also sensitive to the local conformational changes and molecular interactions.

The behavior of proteins is closely related to the protonation states of the residues. Therefore, prediction and measurement of pKa are essential to understand the basic functions of proteins. In this work, we develop a new empirical scheme for protein pKa prediction that is based on deep representation learning. It combines machine learning with atomic environment vector (AEV) and learned quantum mechanical representation from ANI-2x neural network potential (J. Chem. Theory Comput. 2020, 16, 4192). The scheme requires only the coordinate information of a protein as the input and separately estimates the pKa for all five titratable amino acid types. The accuracy of the approach was analyzed with both cross-validation and an external test set of proteins. Obtained results were compared with the widely used empirical approach PROPKA. The new empirical model provides accuracy with MAEs below 0.5 for all amino acid types. It surpasses the accuracy of PROPKA and performs significantly better than the null model. Our model is also sensitive to the local conformational changes and molecular interactions.

Deep generative neural networks have been used increasingly in computational chemistry for de novo design of molecules with desired properties. Many deep learning approaches employ reinforcement learning for optimizing the target properties of the generated molecules. However, the success of this approach is often hampered by the problem of sparse rewards as the majority of the generated molecules are expectedly predicted as inactives. We propose several technical innovations to address this problem and improve the balance between exploration and exploitation modes in reinforcement learning. In a proof-of-concept study, we demonstrate the application of the deep generative recurrent neural network enhanced by several novel technical tricks to designing experimentally validated potent inhibitors of the epidermal growth factor (EGFR). The proposed technical solutions are expected to substantially improve the success rate of finding novel bioactive compounds for specific biological targets using generative and reinforcement learning approaches.

We cover diverse methodologies, computational approaches, and case studies illustrating the ongoing efforts to develop viable drug candidates for treatment of COVID-19.

Abstract High-level quantum mechanical (QM) calculations are indispensable for accurate explanation of natural phenomena on the atomistic level. Their staggering computational cost, however, poses great limitations, which luckily can be lifted to a great extent by exploiting advances in artificial intelligence (AI). Here we introduce the general-purpose, highly transferable artificial intelligence–quantum mechanical method 1 (AIQM1). It approaches the accuracy of the gold-standard coupled cluster QM method with high computational speed of the approximate low-level semiempirical QM methods for the neutral, closed-shell species in the ground state. AIQM1 can provide accurate ground-state energies for diverse organic compounds as well as geometries for even challenging systems such as large conjugated compounds (fullerene C 60 ) close to experiment. This opens an opportunity to investigate chemical compounds with previously unattainable speed and accuracy as we demonstrate by determining geometries of polyyne molecules—the task difficult for both experiment and theory. Noteworthy, our method’s accuracy is also good for ions and excited-state properties, although the neural network part of AIQM1 was never fitted to these properties.

The Hückel Hamiltonian is an incredibly simple tight-binding model known for its ability to capture qualitative physics phenomena arising from electron interactions in molecules and materials. Part of its simplicity arises from using only two types of empirically fit physics-motivated parameters: the first describes the orbital energies on each atom and the second describes electronic interactions and bonding between atoms. By replacing these empirical parameters with machine-learned dynamic values, we vastly increase the accuracy of the extended Hückel model. The dynamic values are generated with a deep neural network, which is trained to reproduce orbital energies and densities derived from density functional theory. The resulting model retains interpretability, while the deep neural network parameterization is smooth and accurate and reproduces insightful features of the original empirical parameterization. Overall, this work shows the promise of utilizing machine learning to formulate simple, accurate, and dynamically parameterized physics models.

Abstract Interatomic potentials derived with Machine Learning algorithms such as Deep-Neural Networks (DNNs), achieve the accuracy of high-fidelity quantum mechanical (QM) methods in areas traditionally dominated by empirical force fields and allow performing massive simulations. Most DNN potentials were parametrized for neutral molecules or closed-shell ions due to architectural limitations. In this work, we propose an improved machine learning framework for simulating open-shell anions and cations. We introduce the AIMNet-NSE (Neural Spin Equilibration) architecture, which can predict molecular energies for an arbitrary combination of molecular charge and spin multiplicity with errors of about 2–3 kcal/mol and spin-charges with error errors ~0.01e for small and medium-sized organic molecules, compared to the reference QM simulations. The AIMNet-NSE model allows to fully bypass QM calculations and derive the ionization potential, electron affinity, and conceptual Density Functional Theory quantities like electronegativity, hardness, and condensed Fukui functions. We show that these descriptors, along with learned atomic representations, could be used to model chemical reactivity through an example of regioselectivity in electrophilic aromatic substitution reactions.

AbstractDespite decades of intensive search for compounds that modulate the activity of particular targets, there are currently small-molecules available only for a small proportion of the human proteome. Effective approaches are therefore required to map the massive space of unexplored compound-target interactions for novel and potent activities. Here, we carried out a crowdsourced benchmarking of predictive models for kinase inhibitor potencies across multiple kinase families using unpublished bioactivity data. The top-performing predictions were based on kernel learning, gradient boosting and deep learning, and their ensemble resulted in predictive accuracy exceeding that of kinase activity assays. We then made new experiments based on the model predictions, which further improved the accuracy of experimental mapping efforts and identified unexpected potencies even for under-studied kinases. The open-source algorithms together with the novel bioactivities between 95 compounds and 295 kinases provide a resource for benchmarking new prediction algorithms and for extending the druggable kinome.

AbstractThe screening of novel materials is an important topic in the field of materials science. Although traditional computational modeling, especially first‐principles approaches, is a very useful and accurate tool to predict the properties of novel materials, it still demands extensive and expensive state‐of‐the‐art computational resources. Additionally, they can often be extremely time consuming. A time and resource efficient machine learning approach to create a dataset of structural properties of 18 million van der Waals layered structures is described. In particular, the authors focus on the interlayer energy and the elastic constant of layered materials composed of two different 2D structures that are important for novel solid lubricant and super‐lubricant materials. It is shown that machine learning models can predict results of computationally expansive approaches (i.e., density functional theory) with high accuracy.

This paper presents TorchANI, a PyTorch based software for training/inferenceof ANI (ANAKIN-ME) deep learning models to obtain potential energy surfaces andother physical properties of molecular systems. ANI is an accurate neural networkpotential originally implemented using C++/CUDA in a program called NeuroChem.Compared with NeuroChem, TorchANI has a design emphasis on being light weight,user friendly, cross platform, and easy to read and modify for fast prototyping, whileallowing acceptable sacrifice on running performance. Because the computation ofatomic environmental vectors (AEVs) and atomic neural networks are all implementedusing PyTorch operators, TorchANI is able to use PyTorch’s autograd engine to automatically compute analytical forces and Hessian matrices, as well as do force trainingwithout additional codes required.

Correction for ‘QSAR without borders’ by Eugene N. Muratov et al., Chem. Soc. Rev., 2020, DOI: 10.1039/d0cs00098a.

Word cloud summary of diverse topics associated with QSAR modeling that are discussed in this review.

We formulate a reaction prediction problem in terms of node-classification in a disconnected graph of source molecules and generalize a graph convolution neural network for disconnected graphs.

AbstractAlchemical free energy methods with molecular mechanics (MM) force fields are now widely used in the prioritization of small molecules for synthesis in structure-enabled drug discovery projects because of their ability to deliver 1–2 kcal mol−1accuracy in well-behaved protein-ligand systems. Surpassing this accuracy limit would significantly reduce the number of compounds that must be synthesized to achieve desired potencies and selectivities in drug design campaigns. However, MM force fields pose a challenge to achieving higher accuracy due to their inability to capture the intricate atomic interactions of the physical systems they model. A major limitation is the accuracy with which ligand intramolecular energetics—especially torsions—can be modeled, as poor modeling of torsional profiles and coupling with other valence degrees of freedom can have a significant impact on binding free energies. Here, we demonstrate how a new generation of hybrid machine learning / molecular mechanics (ML/MM) potentials can deliver significant accuracy improvements in modeling protein-ligand binding affinities. Using a nonequilibrium perturbation approach, we can correct a standard, GPU-accelerated MM alchemical free energy calculation in a simple post-processing step to efficiently recover ML/MM free energies and deliver a significant accuracy improvement with small additional computational effort. To demonstrate the utility of ML/MM free energy calculations, we apply this approach to a benchmark system for predicting kinase:inhibitor binding affinities—a congeneric ligand series for non-receptor tyrosine kinase TYK2 (Tyk2)—wherein state-of-the-art MM free energy calculations (with OPLS2.1) achieve inaccuracies of 0.93±0.12 kcal mol−1in predicting absolute binding free energies. Applying an ML/MM hybrid potential based on the ANI2x ML model and AMBER14SB/TIP3P with the OpenFF 1.0.0 (“Parsley”) small molecule force field as an MM model, we show that it is possible to significantly reduce the error in absolute binding free energies from 0.97 [95% CI: 0.68, 1.21] kcal mol−1(MM) to 0.47 [95% CI: 0.31, 0.63] kcal mol−1(ML/MM).

Abstract Maximum diversification of data is a central theme in building generalized and accurate machine learning (ML) models. In chemistry, ML has been used to develop models for predicting molecular properties, for example quantum mechanics (QM) calculated potential energy surfaces and atomic charge models. The ANI-1x and ANI-1ccx ML-based general-purpose potentials for organic molecules were developed through active learning; an automated data diversification process. Here, we describe the ANI-1x and ANI-1ccx data sets. To demonstrate data diversity, we visualize it with a dimensionality reduction scheme, and contrast against existing data sets. The ANI-1x data set contains multiple QM properties from 5 M density functional theory calculations, while the ANI-1ccx data set contains 500 k data points obtained with an accurate CCSD(T)/CBS extrapolation. Approximately 14 million CPU core-hours were expended to generate this data. Multiple QM calculated properties for the chemical elements C, H, N, and O are provided: energies, atomic forces, multipole moments, atomic charges, etc. We provide this data to the community to aid research and development of ML models for chemistry.

Abstract Motivation Non-ribosomal peptide synthetases (NRPSs) are modular enzymatic machines that catalyze the ribosome-independent production of structurally complex small peptides, many of which have important clinical applications as antibiotics, antifungals and anti-cancer agents. Several groups have tried to expand natural product diversity by intermixing different NRPS modules to create synthetic peptides. This approach has not been as successful as anticipated, suggesting that these modules are not fully interchangeable. Results We explored whether Inter-Modular Linkers (IMLs) impact the ability of NRPS modules to communicate during the synthesis of NRPs. We developed a parser to extract 39 804 IMLs from both well annotated and putative NRPS biosynthetic gene clusters from 39 232 bacterial genomes and established the first IMLs database. We analyzed these IMLs and identified a striking relationship between IMLs and the amino acid substrates of their adjacent modules. More than 92% of the identified IMLs connect modules that activate a particular pair of substrates, suggesting that significant specificity is embedded within these sequences. We therefore propose that incorporating the correct IML is critical when attempting combinatorial biosynthesis of novel NRPS. Availability and implementation The IMLs database as well as the NRPS-Parser have been made available on the web at https://nrps-linker.unc.edu. The entire source code of the project is hosted in GitHub repository (https://github.com/SWFarag/nrps-linker). Supplementary information Supplementary data are available at Bioinformatics online.

Abstract The effectiveness of most cancer targeted therapies is short-lived. Tumors often develop resistance that might be overcome with drug combinations. However, the number of possible combinations is vast, necessitating data-driven approaches to find optimal patient-specific treatments. Here we report AstraZeneca’s large drug combination dataset, consisting of 11,576 experiments from 910 combinations across 85 molecularly characterized cancer cell lines, and results of a DREAM Challenge to evaluate computational strategies for predicting synergistic drug pairs and biomarkers. 160 teams participated to provide a comprehensive methodological development and benchmarking. Winning methods incorporate prior knowledge of drug-target interactions. Synergy is predicted with an accuracy matching biological replicates for >60% of combinations. However, 20% of drug combinations are poorly predicted by all methods. Genomic rationale for synergy predictions are identified, including ADAM17 inhibitor antagonism when combined with PIK3CB/D inhibition contrasting to synergy when combined with other PI3K-pathway inhibitors in PIK3CA mutant cells.

Abstract Computational modeling of chemical and biological systems at atomic resolution is a crucial tool in the chemist’s toolset. The use of computer simulations requires a balance between cost and accuracy: quantum-mechanical methods provide high accuracy but are computationally expensive and scale poorly to large systems, while classical force fields are cheap and scalable, but lack transferability to new systems. Machine learning can be used to achieve the best of both approaches. Here we train a general-purpose neural network potential (ANI-1ccx) that approaches CCSD(T)/CBS accuracy on benchmarks for reaction thermochemistry, isomerization, and drug-like molecular torsions. This is achieved by training a network to DFT data then using transfer learning techniques to retrain on a dataset of gold standard QM calculations (CCSD(T)/CBS) that optimally spans chemical space. The resulting potential is broadly applicable to materials science, biology, and chemistry, and billions of times faster than CCSD(T)/CBS calculations.

AbstractCalculating vibrational properties of crystals using quantum mechanical (QM) methods is a challenging problem in computational material science. This problem is solved using complementary machine learning methods that rapidly and reliably recapitulate entropy, specific heat, effective polycrystalline dielectric function, and a non‐vibrational property (band gap) for materials calculated by accurate but lengthy QM methods. The materials are described mathematically using property‐labeled materials fragment descriptors. The machine learning models predict the QM properties with root mean square errors of 0.31 meV per atom per K for entropy, 0.18 meV per atom per K for specific heat, 4.41 for the trace of the dielectric tensor, and 0.5 eV for band gap. These models are sufficiently accurate to allow rapid screening of large numbers of crystal structures to accelerate material discovery.

Adsorption energies of different nitrogen-containing compounds on two hydroxylated (001) and (100) quartz surfaces are computed.

We introduce a modular, chemically inspired deep neural network model for prediction of several atomic and molecular properties.

We have built a model that ascribes probabilities to the formation of hypothetical compounds, given the proposed oxidation states of the constituent species.

We introduce an artificial intelligence approach to de novo design of molecules with desired physical or biological properties.

The development of accurate and transferable machine learning (ML) potentials for predicting molecular energetics is a challenging task. The process of data generation to train such ML potentials is a task neither well understood nor researched in detail. In this work, we present a fully automated approach for the generation of datasets with the intent of training universal ML potentials. It is based on the concept of active learning (AL) via Query by Committee (QBC), which uses the disagreement between an ensemble of ML potentials to infer the reliability of the ensemble’s prediction. QBC allows the presented AL algorithm to automatically sample regions of chemical space where the ML potential fails to accurately predict the potential energy. AL improves the overall fitness of ANAKIN-ME (ANI) deep learning potentials in rigorous test cases by mitigating human biases in deciding what new training data to use. AL also reduces the training set size to a fraction of the data required when using naive random sampling techniques. To provide validation of our AL approach, we develop the COmprehensive Machine-learning Potential (COMP6) benchmark (publicly available on GitHub) which contains a diverse set of organic molecules. Active learning-based ANI potentials outperform the original random sampled ANI-1 potential with only 10% of the data, while the final active learning-based model vastly outperforms ANI-1 on the COMP6 benchmark after training to only 25% of the data. Finally, we show that our proposed AL technique develops a universal ANI potential (ANI-1x) that provides accurate energy and force predictions on the entire COMP6 benchmark. This universal ML potential achieves a level of accuracy on par with the best ML potentials for single molecules or materials, while remaining applicable to the general class of organic molecules composed of the elements CHNO.

Abstract There are now, in principle, a limitless number of hybrid van der Waals (vdW) heterostructures that can be built from the rapidly growing number of 2D layers. The key question is how to explore this vast parameter space in a practical way. Computational methods can guide experimental work. However, even the most efficient electronic structure methods such as density functional theory, are too time consuming to explore more than a tiny fraction of all possible hybrid 2D materials. A combination of density functional theory (DFT) and machine learning techniques provide a practical method for exploring this parameter space much more efficiently than by DFT or experiments. As a proof of concept, this methodology is applied to predict the interlayer distance and band gap of bilayer heterostructures. The methods quickly and accurately predict these important properties for a large number of hybrid 2D materials. This work paves the way for rapid computational screening of the vast parameter space of vdW heterostructures to identify new hybrid materials with useful and interesting properties.

Atomic and molecular properties could be evaluated from the fundamental Schrodinger’s equation and therefore represent different modalities of the same quantum phenomena. Here we present AIMNet, a modular and chemically inspired deep neural network potential. We used AIMNet with multitarget training to learn multiple modalities of the state of the atom in a molecular system. The resulting model shows on several benchmark datasets the state-of-the-art accuracy, comparable to the results of orders of magnitude more expensive DFT methods. It can simultaneously predict several atomic and molecular properties without an increase in computational cost. With AIMNet we show a new dimension of transferability: the ability to learn new targets utilizing multimodal information from previous training. The model can learn implicit solvation energy (like SMD) utilizing only a fraction of original training data, and archive MAD error of 1.1 kcal/mol compared to experimental solvation free energies in MNSol database.

Atomic and molecular properties could be evaluated from the fundamental Schrodinger’s equation and therefore represent different modalities of the same quantum phenomena. Here we present AIMNet, a modular and chemically inspired deep neural network potential. We used AIMNet with multitarget training to learn multiple modalities of the state of the atom in a molecular system. The resulting model shows on several benchmark datasets the state-of-the-art accuracy, comparable to the results of orders of magnitude more expensive DFT methods. It can simultaneously predict several atomic and molecular properties without an increase in computational cost. With AIMNet we show a new dimension of transferability: the ability to learn new targets utilizing multimodal information from previous training. The model can learn implicit solvation energy (like SMD) utilizing only a fraction of original training data, and archive MAD error of 1.1 kcal/mol compared to experimental solvation free energies in MNSol database.

Atomic and molecular properties could be evaluated from the fundamental Schrodinger’s equation and therefore represent different modalities of the same quantum phenomena. Here we present AIMNet, a modular and chemically inspired deep neural network potential. We used AIMNet with multitarget training to learn multiple modalities of the state of the atom in a molecular system. The resulting model shows on several benchmark datasets the state-of-the-art accuracy, comparable to the results of orders of magnitude more expensive DFT methods. It can simultaneously predict several atomic and molecular properties without an increase in computational cost. With AIMNet we show a new dimension of transferability: the ability to learn new targets utilizing multimodal information from previous training. The model can learn implicit solvation energy (like SMD) utilizing only a fraction of original training data, and archive MAD error of 1.1 kcal/mol compared to experimental solvation free energies in MNSol database.

AbstractAlthough historically materials discovery has been driven by a laborious trial-and-error process, knowledge-driven materials design can now be enabled by the rational combination of Machine Learning methods and materials databases. Here, data from the AFLOW repository for ab initio calculations is combined with Quantitative Materials Structure-Property Relationship models to predict important properties: metal/insulator classification, band gap energy, bulk/shear moduli, Debye temperature and heat capacities. The prediction’s accuracy compares well with the quality of the training data for virtually any stoichiometric inorganic crystalline material, reciprocating the available thermomechanical experimental data. The universality of the approach is attributed to the construction of the descriptors: Property-Labelled Materials Fragments. The representations require only minimal structural input allowing straightforward implementations of simple heuristic design rules.

AbstractOne of the grand challenges in modern theoretical chemistry is designing and implementing approximations that expedite ab initio methods without loss of accuracy. Machine learning (ML) methods are emerging as a powerful approach to constructing various forms of transferable atomistic potentials. They have been successfully applied in a variety of applications in chemistry, biology, catalysis, and solid-state physics. However, these models are heavily dependent on the quality and quantity of data used in their fitting. Fitting highly flexible ML potentials, such as neural networks, comes at a cost: a vast amount of reference data is required to properly train these models. We address this need by providing access to a large computational DFT database, which consists of more than 20 M off equilibrium conformations for 57,462 small organic molecules. We believe it will become a new standard benchmark for comparison of current and future methods in the ML potential community.

We demonstrate how a deep neural network (NN) trained on a data set of quantum mechanical (QM) DFT calculated energies can learn an accurate and transferable atomistic potential for organic molecules containing H, C, N, and O atoms.

AbstractSilicon nanowires (SiNWs) are at the top of the list of materials used in conventional electromechanical devices as well as in strained nanotechnology. Both experimental and theoretical studies showed the size‐dependent character of mechanical properties of SiNWs. However, the surface contaminations, local surface strains, ‘boundary conditions’, native oxide, equipment‐induced errors, and the errors caused by postprocessing of results lead to softening of Young's modulus and extension of the region where the size dependency is seen by experimentalists. Application of improved potentials or advanced theoretical modeling such as inclusion of explicit treatment of temperature and quantum‐mechanical effects allows to show specificity of Young's modulus to the size and shape in case of small (width <4 nm) nanowires. The ductile‐brittle transitions of SiNWs at different temperatures are revealed. Some suggestions on postprocessing techniques are discussed. © 2012 John Wiley & Sons, Ltd.This article is categorized under:Structure and Mechanism > Molecular Structures

AbstractReduction, catalyzed by the bacterial nitroreductases, is the quintessential first step in the biodegradation of a variety of nitroaromatic compounds from contaminated waters and soil. The Enterobacter cloacae nitroreductase (EcNR) enzyme is considered as a prospective biotechnological tool for bioremediation of hazardous nitroaromatic compounds. Using diverse computational methods, we obtain insights into the structural basis of activity and mechanism of its function. We have performed molecular dynamics simulation of EcNR in three different states (free EcNR in oxidized form, fully reduced EcNR with benzoate inhibitor and fully reduced EcNR with nitrobenzene) in explicit solvent and with full electrostatics. Principal Component Analysis (PCA) of the variance‐covariance matrix showed that the complexed nitroreductase becomes more flexible overall upon complexation, particularly helix H6, in the vicinity of the binding site. A multiple sequence alignment was also constructed in order to examine positional constraints on substitution in EcNR. Five regions which are highly conserved within the flavin mononucleotide (FMN) binding site were identified. Obtained results and their implications for EcNR functioning are discussed, and new plausible mechanism has been proposed. Proteins 2012; © 2012 Wiley Periodicals, Inc.

AbstractA number of density functionals was utilized for the calculation of electron attachment free energy for nitrocompounds, quinones and azacyclic compounds. Different solvation models have been tested on the calculation of difference in free energies of solvation of oxidized and reduced forms of nitrocompounds in aqueous solution, quinones in acetonitrile, and azacyclic compounds in dimethylformamide. Gas‐phase free energies evaluated at the mPWB1K/tzvp level and solvation energies obtained using SMD model to compute solvation energies of neutral oxidized forms and PCM(Pauling) to compute solvation energies of anion‐radical reduced forms provide reasonable accuracy of the prediction of electron attachment free energy, difference in free solvation energies of oxidized and reduced forms, and as consequence yield reduction potentials in good agreement with experimental data (mean absolute deviation is 0.15 V). It was also found that SMD/M05‐2X/tzvp method provides reduction potentials with deviation of 0.12 V from the experimental values but in cases of nitrocompounds and quinones this accuracy is achieved due to the cancelation of errors. To predict reduction ability of naturally occurred iron containing species with respect to organic pollutants we exploited experimental data within the framework of Pourbaix (Eh − pH) diagrams. We conclude that surface‐bound Fe(II) as well as certain forms of aqueous Fe(II)aq are capable of reducing a variety of nitroaromatic compounds, quinones and novel high energy materials under basic conditions (pH > 8). At the same time, zero‐valent iron is expected to be active under neutral and acidic conditions. © 2011 Wiley Periodicals, Inc. J Comput Chem, 2011

AbstractReactions of 2‐nitro‐, 4‐nitro‐ and 2,4‐dinitrophenylglycidyl ethers with bicyclo[2.2.1]hept‐5‐ene‐endo‐2‐ylmethylamine in isopropanol have been studied. The mixtures of products were chromatographed on silica gel and eluted with ether or ether/2‐propanol (1:1), the structures of individual products have been confirmed by IR spectra, NMR 1H, 13C spectra, using experiments that involve homonuclear and heteronuclear scalar coupling interactions (COSY, TOCSY, HMQC, HMBC), and mass spectrometry. Amino alcohols as the major products of regioselective aminolysis of epoxides (according to the Krasusky rule) have been obtained. The minor products were the compounds with two hydroxyalkyl fragments at the nitrogen atom. In case of dinitrophenylglycidyl ether, it was the minor product of aryl nucleophilic substitution (SNAr). The abnormal course of aminolysis has been confirmed by the results of quantum‐chemical calculations of activation barries and Free Gibbs energies of the competitive reactions of epoxides (at the B3LYP/6‐311 + G(d,p) level of theory). Copyright © 2010 John Wiley & Sons, Ltd.

AbstractThe theoretical study has been performed to refine the procedure for calculations of Gibbs free energy with a relative accuracy of less than 1 kcal/mol. Three benchmark intermolecular complexes are examined via several quantum‐chemical methods, including the second‐order Moller‐Plesset perturbation (MP2), coupled cluster (CCSD(T)), and density functional (BLYP, B3LYP) theories augmented by Dunnings correlation‐consistent basis sets. The effects of electron correlation, basis set size, and anharmonicity are systematically analyzed, and the results are compared with available experimental data. The results of the calculations suggest that experimental accuracy can be reached only by extrapolation of MP2 and CCSD(T) total energies to the complete basis set. The contribution of anharmonicity to the zero point energy and TΔSint values is fairly small. The new, economic way to reach chemical accuracy in the calculations of the thermodynamic parameters of intermolecular interactions is proposed. In addition, interaction energy (De) and free energy change (ΔA) for considered species have been evaluated by Carr‐Parrinello molecular dynamics (CPMD) simulations and static BLYP‐plane wave calculations. The free energy change along the reaction paths were determined by the thermodynamic integration/”Blue Moon Ensemble” technique. Comparison between obtained values, and available experimental and conventional ab initio results has been made. We found that the accuracy of CPMD simulations is affected by several factors, including statistical uncertainty and convergence of constrained forces (TD integration), and the nature of DFT (density functional theory) functional. The results show that CPMD technique is capable of reproducing interaction and free energy with an accuracy of 1 kcal/mol and 2–3 kcal/mol respectively. © 2007 Wiley Periodicals, Inc. J Comput Chem, 2007