Research publications in machine learning, computational chemistry, and drug discovery from the Isayev Lab at Carnegie Mellon University.https://olexandrisayev.com/en-usolexandr@cmu.edu (Olexandr Isayev)olexandr@cmu.edu (Olexandr Isayev)Copyright 2026 Olexandr IsayevScienceChemistryMachine LearningPublicationshttps://doi.org/10.1021/acs.jctc.5c01794https://doi.org/10.1021/acs.jctc.5c01794Chen, Yuxinxin and Hou, Yi-Fan and Zubatyuk, Roman and Isayev, Olexandr and Dral, Pavlo O.. Journal of Chemical Theory and Computation (2026)Thu, 01 Jan 2026 00:00:00 GMTAIQM3semi-empirical methodscoupled clustermain-group chemistrymachine learning potentialshybrid quantum chemistryChen, Yuxinxin et al.https://doi.org/10.1039/d5dd00380fhttps://doi.org/10.1039/d5dd00380fReidenbach, Danny and Nikitin, Filipp and Isayev, Olexandr and Paliwal, Saee Gopal. Digital Discovery (2026)Thu, 01 Jan 2026 00:00:00 GMT3D molecule generationde novo designmodular co-designgenerative modelsstructure-based designReidenbach, Danny et al.https://doi.org/10.1039/d4sc08572hhttps://doi.org/10.1039/d4sc08572hMachine learned interatomic potentials (MLIPs) are reshaping computational chemistry practices because of their ability to drastically exceed the accuracy-length/time scale tradeoff.Wed, 01 Jan 2025 00:00:00 GMTneutral moleculescharged systemsorganic compoundselemental-organic hybridsuniversal applicabilityAnstine, Dylan M. et al.https://doi.org/10.1021/acs.jcim.5c00395https://doi.org/10.1021/acs.jcim.5c00395Proteochemometric models (PCMs) are used in computational drug discovery to employ both protein and ligand representations jointly for bioactivity prediction. While machine learning (ML) and deep learning (DL) have come to dominate PCMs, often serving as a basis for scoring functions, rigorous evaluation standards have not always been consistently applied. In this study, using kinase-ligand bioactivity prediction as a model system, we highlight the critical roles of data set curation, permutation testing, class imbalances, and various data splitting strategies for mitigating plausible data leakage and embedding quality in determining model performance. Our findings indicate that data splitting and class imbalances are the most critical factors affecting PCM performance, emphasizing the challenges in the generalizing ability of ML/DL-PCMs. We evaluated various protein–ligand descriptors and embeddings, including those augmented with multiple sequence alignment information. However, permutation testing consistently demonstrated that protein embeddings contributed minimally to PCM efficacy. This study advocates for the adoption of stringent evaluation standards to enhance the generalizability of models to out-of-distribution data and improve benchmarking practices.Wed, 01 Jan 2025 00:00:00 GMTproteochemometric modelsmachine learningdeep learningkinase-ligand bioactivity predictiondata curationpermutation testingclass imbalancesdata splittingAvdiunina, Polina et al.https://doi.org/10.1021/acs.jctc.5c01161https://doi.org/10.1021/acs.jctc.5c01161Reaction mechanism search tools have demonstrated the ability to provide insights into likely products and rate-limiting steps of reacting systems. However, reactions involving several concerted bond changes─as can be found in many key steps of natural product syntheses─can complicate the search process. To mitigate these complications, we present a mechanism search strategy particularly suited to help expedite exploration of an exemplary family of such complex reactions, cyclizations. We provide a cost-effective strategy for identifying relevant elementary reaction steps by combining graph-based enumeration schemes and machine learning techniques for intermediate filtering. Key to this approach is our use of a neural network potential (NNP), AIMNet2-rxn, for computational evaluation of each candidate reaction pathway. In this article, we evaluate the NNP's ability to estimate activation energies, demonstrate the correct anticipation of stereoselectivity, and recapitulate complex enabling steps in natural product synthesis.Wed, 01 Jan 2025 00:00:00 GMTreaction pathway explorationgraph-based enumerationintermediate filteringcyclization selectivityaimnet2-rxn applicationCasetti, Nicholas et al.https://doi.org/10.1145/3746252.3761323https://doi.org/10.1145/3746252.3761323Reaction yield prediction underpins computer-aided synthesis prediction (CASP). Formulated as a regression problem that takes both reactants and products as input, this task has been extensively studied using machine learning methods, based on handcrafted fingerprint features, SMILES encoded by Transformers, and molecular graphs encoded by Graph Neural Networks. However, a major limitation of these methods is their inability to effectively capture and model the underlying uncertainties, arising both from the inherently stochastic nature of chemical reaction processes and from inconsistencies or noise in how yields are measured and reported. What makes this seemingly simple regression problem even more challenging is the lack of any principled way to account for the underlying uncertainties, due to missing or unrecorded experimental process (commonly happens in chemical labs). Given these challenges, we propose a new formulation for yield prediction. Rather than assuming a single deterministic yield value for a given reaction, we model the outcome as a probabilistic distribution over three discrete yield regimes: high, medium, and low, reflecting the inherent uncertainty in the reaction process, which is often only partially observed. Accordingly, we propose Proto-Yield, an encoder-agnostic prototype network that models reactions as occurring in one of three yield regimes: high, medium, or low. Without access to full reaction processes, Proto-Yield learns to infer latent regimes and their associated yield distributions from noisy, incomplete training data. During inference, Proto-Yield outputs both a calibrated probability distribution over the yield regimes and the predicted yield conditioned on each regime. Extensive experiments on a 41,000-reaction patent corpus and two high-throughput benchmarks show that Proto-Yield improves R2 by up to 15% and reduces RMSE/MAE by 13% compared to baseline methods.Wed, 01 Jan 2025 00:00:00 GMTyield predictionuncertainty modelingprototype networkprobabilistic distributionschemical reactionsGuo, Kehan et al.https://doi.org/10.1039/d5dd00253bhttps://doi.org/10.1039/d5dd00253bAutonomous experiments are vulnerable to unforeseen adverse events. We developed a transferable ML framework that flags affected HPLC runs in real time and provides expert-level quality control without human oversight.Wed, 01 Jan 2025 00:00:00 GMTautomated quality assurancereal-time monitoringmachine learning algorithmsdata-driven quality controlcloud-based experimentationGusev, Filipp et al.https://doi.org/10.1038/s43588-025-00930-6https://doi.org/10.1038/s43588-025-00930-6As quantum mechanics marks its centennial in 2025, machine learning interatomic potentials have emerged as transformative tools in molecular modeling, bridging quantum mechanical accuracy with classical efficiency. Here we examine their development through four defining challenges-achieving chemical accuracy, maintaining computational efficiency, ensuring interpretability and reaching universal generalizability. We highlight architectural innovations, physics-informed approaches, and foundation models trained on extensive data. Together, these developments chart a path toward predictive, transferable and physically grounded machine learning frameworks for next-generation computational chemistry.Wed, 01 Jan 2025 00:00:00 GMTtransferable potentialsphysics-augmented learningscalable molecular modelingadaptive force fieldsmultiscale modelingKalita, Bhupalee et al.https://doi.org/10.1002/anie.202516763https://doi.org/10.1002/anie.202516763Kalita, Bhupalee and Zubatyuk, Roman and Anstine, Dylan M. and Bergeler, Maike and Settels, Volker and Stork, Conrad and Spicher, Sebastian and Isayev, Olexandr. Angewandte Chemie International Edition (2025)Wed, 01 Jan 2025 00:00:00 GMTAIMNet2-NSEneural network potentialopen-shell chemistryreactive chemistrytransferable potentialradical chemistryKalita, Bhupalee et al.https://doi.org/10.1021/jacsau.5c00667https://doi.org/10.1021/jacsau.5c00667Ring strain energy (RSE) is crucial for understanding molecular reactivity, with broad implications in polymerization, click chemistry, drug discovery and beyond. However, quantitatively determining RSE through experiments or quantum mechanics (QM) is resource-intensive, limiting its application on a large scale. We present a machine learning (ML)-based workflow that enables the reliable and efficient prediction of RSE, entirely bypassing traditional QM calculations. Our workflow employs AIMNet2 machine learning interatomic potentials and Auto3D for the identification of low-energy conformers and RSE computation. Remarkably, it achieves an R 2 of 0.997 and a mean absolute error (MAE) of 0.896 kcal/mol when benchmarked against the ωB97M-D4/Def2-TZVPP method, while running orders of magnitude faster than DFT calculations. To demonstrate the utility of our workflow, we successfully differentiated reactive from nonreactive molecules in copper-free click chemistry, [3 + 2] cycloaddition reaction and ring-opening metathesis polymerization, underscoring its transferability to diverse molecular systems. Additionally, we compiled the RSE Atlas, a computational database encompassing 16,905 single-ring molecules, offering a valuable resource for investigating factors influencing RSE. Our approach transforms RSE into a readily computable property, facilitating its integration into reaction designs.Wed, 01 Jan 2025 00:00:00 GMTconformer identificationstrain energy computationmolecular mechanicshigh-throughput screeningchemical reactivityLiu, Zhen et al.https://doi.org/10.1021/acs.cgd.5c01001https://doi.org/10.1021/acs.cgd.5c01001Identifying thermodynamically stable crystal structures remains a key challenge in materials chemistry. Computational crystal structure prediction (CSP) workflows typically rank candidate structures by lattice energy to assess relative stability. Approaches using self-consistent first-principles calculations become prohibitively expensive, especially when millions of energy evaluations are required for complex molecular systems with many atoms per unit cell. Here, we provide a detailed analysis of our methodology and results from the seventh blind test of crystal structure prediction organized by the Cambridge Crystallographic Data Centre (CCDC). We present an approach that significantly accelerates CSP by training target-specific machine-learned interatomic potentials (MLIPs). AIMNet2 MLIPs are trained on density functional theory (DFT) calculations of molecular clusters, herein referred to as n-mers. We demonstrate that potentials trained on gas phase dispersion-corrected DFT reference data of n-mers successfully extend to crystalline environments, accurately characterizing the CSP landscape and correctly ranking structures by relative stability. Our methodology effectively captures the underlying physics of thermodynamic crystal stability using only molecular cluster data, avoiding the need for expensive periodic calculations. The performance of target-specific AIMNet2 interatomic potentials is illustrated across diverse chemical systems relevant to pharmaceutical, optoelectronic, and agrochemical applications, demonstrating their promise as efficient alternatives to full DFT calculations for routine CSP tasks.Wed, 01 Jan 2025 00:00:00 GMTaimnet2molecular crystalscrystal structure prediction (csp)machine-learned interatomic potentials (mlips)density functional theory (dft)Nayal, Kamal Singh et al.https://doi.org/10.1039/d5dd00206khttps://doi.org/10.1039/d5dd00206kRevisiting GEOM drugs: corrected metrics and novel energy-based structural benchmark enable rigorous evaluation of 3D molecule generative models.Wed, 01 Jan 2025 00:00:00 GMTbenchmarking frameworkmolecular conformation accuracyenergy landscapesmodel validation techniqueschemical structure predictionNikitin, Filipp et al.https://doi.org/10.1002/ange.202513147https://doi.org/10.1002/ange.202513147Rapp, Johann L. and Anstine, Dylan M. and Gusev, Filipp and Nikitin, Filipp and Yun, Kelly H. and Borden, Meredith A. and Bhat, Vittal and Isayev, Olexandr and Leibfarth, Frank A.. Angewandte Chemie (2025)Wed, 01 Jan 2025 00:00:00 GMT3D printingelastomersreinforcement learninghuman-in-the-loopmaterials designsoft materialsRapp, Johann L. et al.https://doi.org/10.1021/acs.jctc.5c00347https://doi.org/10.1021/acs.jctc.5c00347Reactive potentials serve as essential tools for investigating chemical reactions with moderate computational costs. However, traditional reactive potentials often depend on fixed, semiempirical parameters, which limits their accuracy and transferability. Overcoming these limitations can significantly expand the applicability of reactive potentials, enabling the simulation of a broader range of reactions under diverse conditions and the prediction of reaction properties, such as barrier heights. This work introduces ANI-1xBB, a novel ANI-based reactive ML potential trained on off-equilibrium molecular conformers generated through an automated bond-breaking workflow. ANI-1xBB significantly enhances the prediction of reaction energetics, barrier heights, and bond dissociation energies, surpassing those of conventional ANI models. Our results show that ANI-1xBB improves transition state modeling and reaction pathway prediction while generalizing effectively to pericyclic reactions and radical-driven processes. Furthermore, the automated data generation strategy supports the efficient construction of large-scale, high-quality reactive data sets, reducing reliance on expensive QM calculations. This work highlights ANI-1xBB as a practical model for accelerating the development of reactive machine learning potentials, offering new opportunities for modeling reaction phenomena.Wed, 01 Jan 2025 00:00:00 GMToff-equilibrium conformersbond-breaking workflowtransition state modelingpericyclic reactionsradical-driven processesZhang, Shuhao et al.https://doi.org/10.1021/acs.jcim.5c00341https://doi.org/10.1021/acs.jcim.5c00341Machine learning interatomic potentials (MLIPs) have emerged as powerful tools for investigating atomistic systems with high accuracy and a relatively low computational cost. However, a common and unaddressed challenge with many current neural network (NN) MLIP models is their limited ability to accurately predict the relative energies of systems containing isolated or nearly isolated atoms, which appear in various reactive processes. To address this limitation, we present a mathematical technique for modifying any existing atom-centered NN architecture to account for the energies of isolated atoms. The result produces a consistent prediction of the atomization energy (AE) of a system using minimal constraints on the model. Using this technique, we build a model architecture that we call hierarchically interacting particle neural network (HIP-NN)-AE, an AE-constrained version of the HIP-NN, as well as ANI-AE, the AE-constrained version of the accurate NN engine for molecular energies (ANI). Our results demonstrate AE consistency of AE-constrained models, which drastically improves the AE predictions for the models. We compare the AE-constrained approach to unconstrained models as well as models from the literature in other scenarios, such as bond dissociation energies, bond dissociation pathways, and extensibility tests. These results show that the constraints improve the model performance in some of these tasks and do not negatively affect the performance on any tasks. The AE constraint approach thus offers a robust solution to the challenges posed by isolated atoms in energy prediction tasks.Wed, 01 Jan 2025 00:00:00 GMTatomization energy constraintsneural networksreactive chemistrymachine learning interatomic potentialship-nn-aeani-aebond dissociation energiesextensibility testsZhang, Shuhao et al.https://doi.org/10.1039/d5sc04079ehttps://doi.org/10.1039/d5sc04079eRing Vault contains 201 546 cyclic molecules across 11 elements. AIMNet2 with 3D information outperformed 2D models in predicting the electronic properties of cyclic molecules.Wed, 01 Jan 2025 00:00:00 GMThigh-throughput predictionmolecular electronicsscalabilitygeneralizability across elementscomputational efficiencyZheng, Peikun et al.https://doi.org/10.1021/acs.jctc.3c01203https://doi.org/10.1021/acs.jctc.3c01203Dral, Pavlo O. and Ge, Fuchun and Hou, Yi-Fan and Zheng, Peikun and Chen, Yuxinxin and Barbatti, Mario and Isayev, Olexandr and Wang, Cheng and Xue, Bao-Xin and Pinheiro Jr, Max and Su, Yuming and Dai, Yiheng and Chen, Yangtao and Zhang, Lina and Zhang, Shuang and Ullah, Arif and Zhang, Quanhao and Ou, Yanchi. Journal of Chemical Theory and Computation (2024)Mon, 01 Jan 2024 00:00:00 GMTMLatomML platformcomputational chemistry workflowsatomistic simulationsmachine learning interfaceDral, Pavlo O. et al.https://doi.org/10.1039/d3sc06880chttps://doi.org/10.1039/d3sc06880cIn 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.Mon, 01 Jan 2024 00:00:00 GMTmolecular dynamicsvirtual screeningbinding affinitydocking algorithmscomputational drug designGutkin, Evgeny et al.https://doi.org/10.1021/acs.jctc.4c01052https://doi.org/10.1021/acs.jctc.4c01052Deep learning Neural Networks (NN) have been developed in the field of molecular modeling for the purpose of circumventing the high computational cost of quantum-mechanical calculations while rivaling their accuracies. Although these networks have found great success, they generally lack the ability to accurately describe long-range interactions, which makes them unusable for extended molecular systems. Herein, we provide a method for partially retraining the deep learning general-use neural network ANI, in which the long-range interactions are represented via atomic electrostatic potentials. The electrostatic potentials, generated with polarizable effective fragment potentials (EFP), are used as an additional input feature for the network. This new ANI/EFP network can predict solute-solvent interaction energies on a trained data set with a kcal/mol accuracy. It also shows promise in predicting the interaction energies of a solute in solvent environments that have not been included in a training data set. The proposed protocol can be taken as an example and further developed, leading to highly accurate and transferable neural network potentials capable of handling long-range interactions and extended molecular systems.Mon, 01 Jan 2024 00:00:00 GMTelectrostatic potential enhancementani model extensionlong-range interaction correctionsolute-solvent energy predictiontransferable neural networkHaghiri, Shahed et al.https://doi.org/10.1021/jacs.4c05245https://doi.org/10.1021/jacs.4c05245Crystalline organic semiconductors are known to have improved charge carrier mobility and exciton diffusion length in comparison to their amorphous counterparts. Certain organic molecular thin films can be transitioned from initially prepared amorphous layers to large-scale crystalline films via abrupt thermal annealing. Ideally, these films crystallize as platelets with long-range-ordered domains on the scale of tens to hundreds of microns. However, other organic molecular thin films may instead crystallize as spherulites or resist crystallization entirely. Organic molecules that have the capability of transforming into a platelet morphology feature both high melting point (Tm) and crystallization driving force (ΔGc). In this work, we employed machine learning (ML) to identify candidate organic materials with the potential to crystallize into platelets by estimating the aforementioned thermal properties. Six organic molecules identified by the ML algorithm were experimentally evaluated; three crystallized as platelets, one crystallized as a spherulite, and two resisted thin film crystallization. These results demonstrate a successful application of ML in the scope of predicting thermal properties of organic molecules and reinforce the principles of Tm and ΔGc as metrics that aid in predicting the crystallization behavior of organic thin films.Mon, 01 Jan 2024 00:00:00 GMTcrystal structure predictionmolecular designthermodynamics of crystallizationorganic electronicsproperty prediction modelsJohnson, Holly M. et al.https://doi.org/10.1039/d3dd00210ahttps://doi.org/10.1039/d3dd00210aThe RRCGAN, validated through DFT, demonstrates success in generating chemically valid molecules targeting energy gap values with 75% of the generated molecules have RE of &lt;20% of the targeted values.Mon, 01 Jan 2024 00:00:00 GMTdeep learningbiasing propertiestransfer learningmolecular generationproperty optimizationSattari, Kianoosh et al.https://doi.org/10.1038/s41573-023-00832-0https://doi.org/10.1038/s41573-023-00832-0Quantitative structure-activity relationship (QSAR) modelling, an approach that was introduced 60 years ago, is widely used in computer-aided drug design. In recent years, progress in artificial intelligence techniques, such as deep learning, the rapid growth of databases of molecules for virtual screening and dramatic improvements in computational power have supported the emergence of a new field of QSAR applications that we term 'deep QSAR'. Marking a decade from the pioneering applications of deep QSAR to tasks involved in small-molecule drug discovery, we herein describe key advances in the field, including deep generative and reinforcement learning approaches in molecular design, deep learning models for synthetic planning and the application of deep QSAR models in structure-based virtual screening. We also reflect on the emergence of quantum computing, which promises to further accelerate deep QSAR applications and the need for open-source and democratized resources to support computer-aided drug design.Mon, 01 Jan 2024 00:00:00 GMTneural networksmolecular designdeep reinforcement learningqsar applicationscomputational powerTropsha, Alexander et al.https://doi.org/10.1038/s41557-023-01427-3https://doi.org/10.1038/s41557-023-01427-3Abstract 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.Mon, 01 Jan 2024 00:00:00 GMTautomated samplingcondensed-phase chemistryhigh-throughput experimentationani-1xnrmachine learning force fieldsZhang, Shuhao et al.https://doi.org/10.1021/jacs.2c13467https://doi.org/10.1021/jacs.2c13467Traditional computational approaches to design chemical species are limited by the need to compute properties for a vast number of candidates, e.g., by discriminative modeling. Therefore, inverse design methods aim to start from the desired property and optimize a corresponding chemical structure. From a machine learning viewpoint, the inverse design problem can be addressed through so-called generative modeling. Mathematically, discriminative models are defined by learning the probability distribution function of properties given the molecular or material structure. In contrast, a generative model seeks to exploit the joint probability of a chemical species with target characteristics. The overarching idea of generative modeling is to implement a system that produces novel compounds that are expected to have a desired set of chemical features, effectively sidestepping issues found in the forward design process. In this contribution, we overview and critically analyze popular generative algorithms like generative adversarial networks, variational autoencoders, flow, and diffusion models. We highlight key differences between each of the models, provide insights into recent success stories, and discuss outstanding challenges for realizing generative modeling discovered solutions in chemical applications.Sun, 01 Jan 2023 00:00:00 GMTgenerative-modelinginverse-design-methodsmolecular-generationvirtual-screeningadmet-propertiesAnstine, Dylan M. et al.https://doi.org/10.1021/acs.jpca.2c06778https://doi.org/10.1021/acs.jpca.2c06778Advances in machine learned interatomic potentials (MLIPs), such as those using neural networks, have resulted in short-range models that can infer interaction energies with near ab initio accuracy and orders of magnitude reduced computational cost. For many atom systems, including macromolecules, biomolecules, and condensed matter, model accuracy can become reliant on the description of short- and long-range physical interactions. The latter terms can be difficult to incorporate into an MLIP framework. Recent research has produced numerous models with considerations for nonlocal electrostatic and dispersion interactions, leading to a large range of applications that can be addressed using MLIPs. In light of this, we present a Perspective focused on key methodologies and models being used where the presence of nonlocal physics and chemistry are crucial for describing system properties. The strategies covered include MLIPs augmented with dispersion corrections, electrostatics calculated with charges predicted from atomic environment descriptors, the use of self-consistency and message passing iterations to propagated nonlocal system information, and charges obtained via equilibration schemes. We aim to provide a pointed discussion to support the development of machine learning-based interatomic potentials for systems where contributions from only nearsighted terms are deficient.Sun, 01 Jan 2023 00:00:00 GMTnonlocal physicsself-consistencymessage passingequilibration schemeslong-range interactionsAnstine, Dylan M. et al.https://doi.org/10.1039/d3cp90129ghttps://doi.org/10.1039/d3cp90129gThis themed collection includes a collection of articles on Insightful Machine Learning for Physical Chemistry.Sun, 01 Jan 2023 00:00:00 GMTneural networksforce fieldsfeature engineeringhyperparameter optimizationbenchmarkingClark, Aurora E. et al.https://doi.org/10.1063/5.0151833https://doi.org/10.1063/5.0151833Fedik, Nikita and Nebgen, Benjamin and Lubbers, Nicholas and Barros, Kipton and Kulichenko, Maksim and Li, Ying Wai and Zubatyuk, Roman and Messerly, Richard and Isayev, Olexandr and Tretiak, Sergei. The Journal of Chemical Physics (2023)Sun, 01 Jan 2023 00:00:00 GMTsemi-empirical methodsmachine learninghybrid quantum chemistrycomputational chemistrymethod synergyFedik, Nikita et al.https://doi.org/10.1021/acs.jcim.2c01052https://doi.org/10.1021/acs.jcim.2c01052In silico identification of potent protein inhibitors commonly requires prediction of a ligand binding free energy (BFE). Thermodynamics integration (TI) based on molecular dynamics (MD) simulations is a BFE calculation method capable of acquiring accurate BFE, but it is computationally expensive and time-consuming. In this work, we have developed an efficient automated workflow for identifying compounds with the lowest BFE among thousands of congeneric ligands, which requires only hundreds of TI calculations. Automated machine learning (AutoML) orchestrated by active learning (AL) in an AL-AutoML workflow allows unbiased and efficient search for a small set of best-performing molecules. We have applied this workflow to select inhibitors of the SARS-CoV-2 papain-like protease and were able to find 133 compounds with improved binding affinity, including 16 compounds with better than 100-fold binding affinity improvement. We obtained a hit rate that outperforms that expected of traditional expert medicinal chemist-guided campaigns. Thus, we demonstrate that the combination of AL and AutoML with free energy simulations provides at least 20× speedup relative to the naïve brute force approaches.Sun, 01 Jan 2023 00:00:00 GMTmachine learning workflowsefficient optimizationrelative binding affinitylead optimizationalgorithmic drug designGusev, Filipp et al.https://doi.org/10.1039/d2sc04815ahttps://doi.org/10.1039/d2sc04815aJaffrelot Inizan, Théo and Plé, Thomas and Adjoua, Olivier and Ren, Pengyu and Gökcan, Hatice and Isayev, Olexandr and Lagardère, Louis and Piquemal, Jean-Philip. Chemical Science (2023)Sun, 01 Jan 2023 00:00:00 GMTdeep neural networkspolarizable potentialsbiomolecular simulationslong-range interactionshybrid potentialsscalabilityJaffrelot Inizan, Théo et al.https://doi.org/10.1039/d3sc03902ahttps://doi.org/10.1039/d3sc03902aLiu, Zhen and Moroz, Yurii S. and Isayev, Olexandr. Chemical Science (2023)Sun, 01 Jan 2023 00:00:00 GMTamide couplingyield predictionmodel robustnessreaction predictionbenchmarkingsynthesis planningLiu, Zhen et al.https://doi.org/10.1021/acs.jpcc.3c02384https://doi.org/10.1021/acs.jpcc.3c02384[This retracts the article DOI: 10.1021/acsanm.4c06844.].Sun, 01 Jan 2023 00:00:00 GMTepitaxial interfacestcnq-ttf interfacecrystal structure predictionogre softwaremachine learning potentialsMoayedpour, Saeed et al.https://doi.org/10.1039/d3sc02408chttps://doi.org/10.1039/d3sc02408cZhao, Qiyuan and Anstine, Dylan M. and Isayev, Olexandr and Savoie, Brett M.. Chemical Science (2023)Sun, 01 Jan 2023 00:00:00 GMTdelta machine learningreaction property predictiontransfer learningtransition statesbarrier heightsZhao, Qiyuan et al.https://doi.org/10.1038/s41570-022-00416-3https://doi.org/10.1038/s41570-022-00416-3Fedik, Nikita and Zubatyuk, Roman and Kulichenko, Maksim and Lubbers, Nicholas and Smith, Justin S. and Nebgen, Benjamin and Messerly, Richard and Li, Ying Wai and Boldyrev, Alexander I. and Barros, Kipton and Isayev, Olexandr and Tretiak, Sergei. Nature Reviews Chemistry (2022)Sat, 01 Jan 2022 00:00:00 GMTmachine learning potentialsmolecular propertiesreviewatomistic methodsneural networksproperty predictionFedik, Nikita et al.https://doi.org/10.1021/acs.jcim.2c00630https://doi.org/10.1021/acs.jcim.2c00630Pyruvate dehydrogenase complex (PDC) deficiency is a major cause of primary lactic acidemia resulting in high morbidity and mortality, with limited therapeutic options. The E1 component of the mitochondrial multienzyme PDC (PDC-E1) is a symmetric dimer of heterodimers (&#x3b1;&#x3b2;/&#x3b1;'&#x3b2;') encoded by the <i>PDHA1</i> and <i>PDHB</i> genes, with two symmetric active sites each consisting of highly conserved phosphorylation loops A and B. <i>PDHA1</i> mutations are responsible for 82-88% of cases. Greater than 85% of E1&#x3b1; residues with disease-causing missense mutations (DMMs) are solvent-inaccessible, with &#x223c;30% among those involved in subunit-subunit interface contact (SSIC). We performed molecular dynamics simulations of wild-type (WT) PDC-E1 and E1 variants with E1&#x3b1; DMMs at R349 and W185 (residues involved in SSIC), to investigate their impact on human PDC-E1 structure. We evaluated the change in E1 structure and dynamics and examined their implications on E1 function with the specific DMMs. We found that the dynamics of phosphorylation Loop A, which is crucial for E1 biological activity, changes with DMMs that are at least about 15 &#xc5; away. Because communication is essential for PDC-E1 activity (with alternating active sites), we also investigated the possible communication network within WT PDC-E1 via centrality analysis. We observed that DMMs altered/disrupted the communication network of PDC-E1. Collectively, these results indicate allosteric effect in PDC-E1, with implications for the development of novel small-molecule therapeutics for specific recurrent E1&#x3b1; DMMs such as replacements of R349 responsible for &#x223c;10% of PDC deficiency due to E1&#x3b1; DMMs.Sat, 01 Jan 2022 00:00:00 GMTmolecular dynamics simulationse1α variantspdc deficiencyallosteric effectscommunication networkssmall-molecule therapeuticsGokcan, Hatice et al.https://doi.org/10.1039/d1sc05610ghttps://doi.org/10.1039/d1sc05610gWe developed new empirical ML model for protein pKaprediction with MAEs below 0.5 for all amino acid types.Sat, 01 Jan 2022 00:00:00 GMTrepresentation learningprotein pka predictionempirical modelsmachine learning in biochemistryamino acid analysisGokcan, Hatice et al.https://doi.org/10.1038/s42004-022-00733-0https://doi.org/10.1038/s42004-022-00733-0Korshunova, Maria and Huang, Niles and Capuzzi, Stephen and Radchenko, Dmytro S. and Savych, Olena and Moroz, Yuriy S. and Wells, Carrow I. and Willson, Timothy M. and Tropsha, Alexander and Isayev, Olexandr. Communications Chemistry (2022)Sat, 01 Jan 2022 00:00:00 GMTgenerative modelsreinforcement learningde novo drug designbioactive compoundsvirtual screeningautomated designKorshunova, Maria et al.https://doi.org/10.1021/acs.jcim.2c00817https://doi.org/10.1021/acs.jcim.2c00817Computational programs accelerate the chemical discovery processes but often need proper three-dimensional molecular information as part of the input. Getting optimal molecular structures is challenging because it requires enumerating and optimizing a huge space of stereoisomers and conformers. We developed the Python-based Auto3D package for generating the low-energy 3D structures using SMILES as the input. Auto3D is based on state-of-the-art algorithms and can automatize the isomer enumeration and duplicate filtering process, 3D building process, geometry optimization, and ranking process. Tested on 50 molecules with multiple unspecified stereocenters, Auto3D is guaranteed to find the stereoconfiguration that yields the lowest-energy conformer. With Auto3D, we provide an extension of the ANI model. The new model, dubbed ANI-2xt, is trained on a tautomer-rich data set. ANI-2xt is benchmarked with DFT methods on geometry optimization and electronic and Gibbs free energy calculations. Compared with ANI-2x, ANI-2xt provides a 42% error reduction for tautomeric reaction energy calculations when using the gold-standard coupled-cluster calculation as the reference. ANI-2xt can accurately predict the energies and is several orders of magnitude faster than DFT methods.Sat, 01 Jan 2022 00:00:00 GMTgeometry optimizationtautomer-rich datasetstereoconfiguration predictionmachine learning accelerationenergy landscape explorationLiu, Zhen et al.https://doi.org/10.1038/s42256-022-00463-xhttps://doi.org/10.1038/s42256-022-00463-xThere has been a recent focus within Alzheimer's Disease (AD) research on potential CSF biomarkers as diagnostic tools. PET imaging and CSF biomarkers have been used to determine the amyloid status of individuals when characterising AD. While these measures have largely correlated with each other, it is also suggested that there could be patients who can have discordant results. However, it is still unclear what proportion of AD subjects demonstrate the discordance. Additionally, whether these changes are associated with different pathological and clinical characteristics of the patients. Here, we investigated the discordance between CSF A&#x3b2;42:40 and amyloid status and evaluated whether there are any significant changes in their levels of tau aggregation. 313 participants' data were selected from the ADNI database. The cutoff for A&#x3b2;42:40 was then calculated using the Youden Index resulting from receiver operating characteristic (ROC) curve. The formula used was J&#xa0;=&#xa0;maxc Se (c) + Sp (c) - 1, resulting in a maximum cutoff of J&#xa0;=&#xa0;0.057 for A&#x3b2;42:40. This cutoff produced a discordance rate of 10.5%, with n&#xa0;=&#xa0;33 of 313 patients being discordant. We then performed t-tests to investigate the potential clinical differences between concordant and discordant patients. Independent samples t-tests indicated that levels of pTau181 and total Tau significantly differed between the concordant and discordant groups. pTau181 was significantly lower in the discordant group (t(71.2)=-3.166, p&#xa0;=&#xa0;.002), as was Tau (t(53.9)=-2.046, p&#xa0;=&#xa0;.046). Further data is found in Figure 1. Demonstration of different levels of tau deposition in the discordant groups implies there may be other pathological processes influencing neurodegeneration in these subjects. A discordance of 10% between CSF amyloid and PET amyloid implies that we should evaluate these subjects in greater detail before enrolling them in intervention studies. References 1. Hansson, O. et al. (2019). https://doi.org/10.1186/s13195-019-0485-0 2. Pyun, JM. et al. (2024) https://doi.org/10.1038/s41398-024-02766-6.Sat, 01 Jan 2022 00:00:00 GMTgpu computingdeep learningalzheimer's diseasecsf biomarkersdiagnostic toolsPandey, Mohit et al.https://doi.org/10.1021/acs.jpclett.2c00734https://doi.org/10.1021/acs.jpclett.2c00734Zheng, Peikun and Yang, Wudi and Wu, Wei and Isayev, Olexandr and Dral, Pavlo O.. The Journal of Physical Chemistry Letters (2022)Sat, 01 Jan 2022 00:00:00 GMTenthalpy of formationchemical accuracythermochemistrydata-driven methodsgeneral-purpose modelsZheng, Peikun et al.https://doi.org/10.1038/s41557-021-00716-zhttps://doi.org/10.1038/s41557-021-00716-zArtrith, Nongnuch and Butler, Keith T. and Coudert, François-Xavier and Han, Seungwu and Isayev, Olexandr and Jain, Anubhav and Walsh, Aron. Nature Chemistry (2021)Fri, 01 Jan 2021 00:00:00 GMTbest practicesmachine learning workflowsneural network applicationsmethodology guidelinesinterdisciplinary approachesArtrith, Nongnuch et al.https://doi.org/10.1038/s41467-021-23165-1https://doi.org/10.1038/s41467-021-23165-1Cichońska, Anna and Ravikumar, Balaguru and Allaway, Robert J. and Wan, Fangping and Park, Sungjoon and Isayev, Olexandr and Li, Shuya and Mason, Michael and Lamb, Andrew and Tanoli, Ziaurrehman and Jeon, Minji and Kim, Sunkyu and Popova, Mariya and Capuzzi, Stephen and Zeng, Jianyang and Dang, Kristen and Koytiger, Gregory and Kang, Jaewoo and Wells, Carrow I. and Willson, Timothy M. and Tan, Mehmet and Huang, Chih-Han and Shih, Edward S. C. and Chen, Tsai-Min and Wu, Chih-Hsun and Fang, Wei-Quan and Chen, Jhih-Yu and Hwang, Ming-Jing and Wang, Xiaokang and Ben Guebila, Marouen and Shamsaei, Behrouz and Singh, Sourav and Nguyen, Thin and Karimi, Mostafa and Wu, Di and Wang, Zhangyang and Shen, Yang and Öztürk, Hakime and Ozkirimli, Elif and Özgür, Arzucan and Lim, Hansaim and Xie, Lei and Kanev, Georgi K. and Kooistra, Albert J. and Westerman, Bart A. and Terzopoulos, Panagiotis and Ntagiantas, Konstantinos and Fotis, Christos and Alexopoulos, Leonidas and Boeckaerts, Dimitri and Stock, Michiel and De Baets, Bernard and Briers, Yves and Luo, Yunan and Hu, Hailin and Peng, Jian and Dogan, Tunca and Rifaioglu, Ahmet S. and Atas, Heval and Atalay, Rengul Cetin and Atalay, Volkan and Martin, Maria J. and Jeon, Minji and Lee, Junhyun and Yun, Seongjun and Kim, Bumsoo and Chang, Buru and Turu, Gábor and Misák, Ádám and Szalai, Bence and Hunyady, László and Lienhard, Matthias and Prasse, Paul and Bachmann, Ivo and Ganzlin, Julia and Barel, Gal and Herwig, Ralf and Oršolić, Davor and Lučić, Bono and Stepanić, Višnja and Šmuc, Tomislav and Oprea, Tudor I. and Schlessinger, Avner and Drewry, David H. and Stolovitzky, Gustavo and Wennerberg, Krister and Guinney, Justin and Aittokallio, Tero. Nature Communications (2021)Fri, 01 Jan 2021 00:00:00 GMTkinase inhibitorsvirtual screeningtarget spacecrowdsourcingchemical biologyselectivityCichońska, Anna et al.https://doi.org/10.1109/jiot.2021.3073904https://doi.org/10.1109/jiot.2021.3073904As COVID-19 hounds the world, the common cause of finding a swift solution to manage the pandemic has brought together researchers, institutions, governments, and society at large. The Internet of Things (IoT), artificial intelligence (AI)—including machine learning (ML) and Big Data analytics—as well as Robotics and Blockchain, are the four decisive areas of technological innovation that have been ingenuity harnessed to fight this pandemic and future ones. While these highly interrelated smart and connected health technologies cannot resolve the pandemic overnight and may not be the only answer to the crisis, they can provide greater insight into the disease and support frontline efforts to prevent and control the pandemic. This article provides a blend of discussions on the contribution of these digital technologies, propose several complementary and multidisciplinary techniques to combat COVID-19, offer opportunities for more holistic studies, and accelerate knowledge acquisition and scientific discoveries in pandemic research. First, four areas, where IoT can contribute are discussed, namely: 1) tracking and tracing; 2) remote patient monitoring (RPM) by wearable IoT (WIoT); 3) personal digital twins (PDTs); and 4) real-life use case: ICT/IoT solution in South Korea. Second, the role and novel applications of AI are explained, namely: 1) diagnosis and prognosis; 2) risk prediction; 3) vaccine and drug development; 4) research data set; 5) early warnings and alerts; 6) social control and fake news detection; and 7) communication and chatbot. Third, the main uses of robotics and drone technology are analyzed, including: 1) crowd surveillance; 2) public announcements; 3) screening and diagnosis; and 4) essential supply delivery. Finally, we discuss how distributed ledger technologies (DLTs), of which blockchain is a common example, can be combined with other technologies for tackling COVID-19.Fri, 01 Jan 2021 00:00:00 GMTcovid-19iotaidrug developmentvaccine discoveryblockchainFirouzi, Farshad et al.https://doi.org/10.1002/aisy.202100080https://doi.org/10.1002/aisy.202100080Fronzi, Marco and Isayev, Olexandr and Winkler, David A. and Shapter, Joseph G. and Ellis, Amanda V. and Sherrell, Peter C. and Shepelin, Nick A. and Corletto, Alexander and Ford, Michael J.. Advanced Intelligent Systems (2021)Fri, 01 Jan 2021 00:00:00 GMTBayesian neural networksactive learningbandgap predictionvan der Waals heterostructures2D materialsFronzi, Marco et al.https://doi.org/10.1002/wcms.1564https://doi.org/10.1002/wcms.1564AbstractThe 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 &gt; Artificial Intelligence/Machine LearningMolecular and Statistical Mechanics &gt; Molecular InteractionsSoftware &gt; Molecular ModelingFri, 01 Jan 2021 00:00:00 GMTneural network potentialsmachine learningmolecular mechanicsdrug discoverycomputational chemistryGokcan, Hatice et al.https://doi.org/10.1021/acs.jcim.0c00971https://doi.org/10.1021/acs.jcim.0c00971Deep learning models have demonstrated outstanding results in many data-rich areas of research, such as computer vision and natural language processing. Currently, there is a rise of deep learning in computational chemistry and materials informatics, where deep learning could be effectively applied in modeling the relationship between chemical structures and their properties. With the immense growth of chemical and materials data, deep learning models can begin to outperform conventional machine learning techniques such as random forest, support vector machines, and nearest neighbor. Herein, we introduce OpenChem, a PyTorch-based deep learning toolkit for computational chemistry and drug design. OpenChem offers easy and fast model development, modular software design, and several data preprocessing modules. It is freely available via the GitHub repository.Fri, 01 Jan 2021 00:00:00 GMTneural network modelsmodular designdata augmentationhyperparameter tuningopen-source toolkitKorshunova, Maria et al.https://doi.org/10.1039/d0cs01065khttps://doi.org/10.1039/d0cs01065kWe cover diverse methodologies, computational approaches, and case studies illustrating the ongoing efforts to develop viable drug candidates for treatment of COVID-19.Fri, 01 Jan 2021 00:00:00 GMTcomputational drug designcovid-19 therapeuticsvirtual screening strategiesmethodology benchmarkingdrug development pipelineMuratov, Eugene N. et al.https://doi.org/10.1021/jacs.1c08181https://doi.org/10.1021/jacs.1c08181Modern polymer science suffers from the curse of multidimensionality. The large chemical space imposed by including combinations of monomers into a statistical copolymer overwhelms polymer synthesis and characterization technology and limits the ability to systematically study structure-property relationships. To tackle this challenge in the context of ¹⁹F magnetic resonance imaging (MRI) agents, we pursued a computer-guided materials discovery approach that combines synergistic innovations in automated flow synthesis and machine learning (ML) method development. A software-controlled, continuous polymer synthesis platform was developed to enable iterative experimental-computational cycles that resulted in the synthesis of 397 unique copolymer compositions within a six-variable compositional space. The nonintuitive design criteria identified by ML, which were accomplished by exploring <0.9% of the overall compositional space, lead to the identification of >10 copolymer compositions that outperformed state-of-the-art materials.Fri, 01 Jan 2021 00:00:00 GMTiterative experimental-computational cycleshigh-throughput screeningnonintuitive design criteriastate-of-the-art materials comparisonpolymer informaticsReis, Marcus et al.https://doi.org/10.1038/s41467-021-27340-2https://doi.org/10.1038/s41467-021-27340-2Zheng, Peikun and Zubatyuk, Roman and Wu, Wei and Isayev, Olexandr and Dral, Pavlo O.. Nature Communications (2021)Fri, 01 Jan 2021 00:00:00 GMTartificial intelligencequantum chemistrysemi-empirical methodsbroad applicabilityhybrid methodsAIQMZheng, Peikun et al.https://doi.org/10.1063/5.0052857https://doi.org/10.1063/5.0052857Zubatiuk, Tetiana and Nebgen, Benjamin and Lubbers, Nicholas and Smith, Justin S. and Zubatyuk, Roman and Zhou, Guoqing and Koh, Christopher and Barros, Kipton and Isayev, Olexandr and Tretiak, Sergei. The Journal of Chemical Physics (2021)Fri, 01 Jan 2021 00:00:00 GMTHuckel theorydeep learningsemi-empiricalphysics-informedelectronic structureZubatiuk, Tetiana et al.https://doi.org/10.1021/acs.accounts.0c00868https://doi.org/10.1021/acs.accounts.0c00868ConspectusMachine learning interatomic potentials (MLIPs) are widely used for describing molecular energy and continue bridging the speed and accuracy gap between quantum mechanical (QM) and classical approaches like force fields. In this Account, we focus on the out-of-the-box approaches to developing transferable MLIPs for diverse chemical tasks. First, we introduce the "Accurate Neural Network engine for Molecular Energies," ANAKIN-ME, method (or ANI for short). The ANI model utilizes Justin Smith Symmetry Functions (JSSFs) and realizes training for vast data sets. The training data set of several orders of magnitude larger than before has become the key factor of the knowledge transferability and flexibility of MLIPs. As the quantity, quality, and types of interactions included in the training data set will dictate the accuracy of MLIPs, the task of proper data selection and model training could be assisted with advanced methods like active learning (AL), transfer learning (TL), and multitask learning (MTL).Next, we describe the AIMNet "Atoms-in-Molecules Network" that was inspired by the quantum theory of atoms in molecules. The AIMNet architecture lifts multiple limitations in MLIPs. It encodes long-range interactions and learnable representations of chemical elements. We also discuss the AIMNet-ME model that expands the applicability domain of AIMNet from neutral molecules toward open-shell systems. The AIMNet-ME encompasses a dependence of the potential on molecular charge and spin. It brings ML and physical models one step closer, ensuring the correct molecular energy behavior over the total molecular charge.We finally describe perhaps the simplest possible physics-aware model, which combines ML and the extended Hückel method. In ML-EHM, "Hierarchically Interacting Particle Neural Network," HIP-NN generates the set of a molecule- and environment-dependent Hamiltonian elements αμμ and K‡. As a test example, we show how in contrast to traditional Hückel theory, ML-EHM correctly describes orbital crossing with bond rotations. Hence it learns the underlying physics, highlighting that the inclusion of proper physical constraints and symmetries could significantly improve ML model generalization.Fri, 01 Jan 2021 00:00:00 GMTmachine learning potentialsaniaimnettransfer learningactive learningphysics-aware aiextended hückel methodmolecular energyZubatiuk, Tetiana et al.