ZeoSyn: A Comprehensive Zeolite Synthesis Dataset Enabling Machine-Learning Rationalization of Hydrothermal Parameters.

Zeolites, nanoporous aluminosilicates with well-defined porous structures, are versatile materials with applications in catalysis, gas separation, and ion exchange. Hydrothermal synthesis is widely used for zeolite production, offering control over composition, crystallinity, and pore size. However, the intricate interplay of synthesis parameters necessitates a comprehensive understanding of synthesis-structure relationships to optimize the synthesis process. Hitherto, public zeolite synthesis databases only contain a subset of parameters and are small in scale, comprising up to a few thousand synthesis routes. We present ZeoSyn, a dataset of 23,961 zeolite hydrothermal synthesis routes, encompassing 233 zeolite topologies and 921 organic structure-directing agents (OSDAs). Each synthesis route comprises comprehensive synthesis parameters: 1) gel composition, 2) reaction conditions, 3) OSDAs, and 4) zeolite products. Using ZeoSyn, we develop a machine learning classifier to predict the resultant zeolite given a synthesis route with >70% accuracy. We employ SHapley Additive exPlanations (SHAP) to uncover key synthesis parameters for >200 zeolite frameworks. We introduce an aggregation approach to extend SHAP to all building units. We demonstrate applications of this approach to phase-selective and intergrowth synthesis. This comprehensive analysis illuminates the synthesis parameters pivotal in driving zeolite crystallization, offering the potential to guide the synthesis of desired zeolites. The dataset is available at https://github.com/eltonpan/zeosyn_dataset.

Design of Cytotoxic T Cell Epitopes by Machine Learning of Human Degrons.

Antigen processing is critical for therapeutic vaccines to generate epitopes for priming cytotoxic T cell responses against cancer and pathogens, but insufficient processing often limits the quantity of epitopes released. We address this challenge using machine learning to ascribe a proteasomal degradation score to epitope sequences. Epitopes with varying scores were translocated into cells using nontoxic anthrax proteins. Epitopes with a low score show pronounced immunogenicity due to antigen processing, but epitopes with a high score show limited immunogenicity. This work sheds light on the sequence-activity relationships between proteasomal degradation and epitope immunogenicity. We anticipate that future efforts to incorporate proteasomal degradation signals into vaccine designs will lead to enhanced cytotoxic T cell priming by these vaccines in clinical settings.

One-Pot Synthesis of CHA/ERI-Type Zeolite Intergrowth from a Single Multiselective Organic Structure-Directing Agent.

We report the one-pot synthesis of a chabazite (CHA)/erionite (ERI)-type zeolite intergrowth structure characterized by adjustable extents of intergrowth enrichment and Si/Al molar ratios. This method utilizes readily synthesizable 6-azaspiro[5.6]dodecan-6-ium as the exclusive organic structure-directing agent (OSDA) within a potassium-dominant environment. High-throughput simulations were used to accurately determine the templating energy and molecular shape, facilitating the selection of an optimally biselective OSDA from among thousands of prospective candidates. The coexistence of the crystal phases, forming a distinct structure comprising disk-like CHA regions bridged by ERI-rich pillars, was corroborated via rigorous powder X-ray diffraction and integrated differential-phase contrast scanning transmission electron microscopy (iDPC S/TEM) analyses. iDPC S/TEM imaging further revealed the presence of single offretite layers dispersed within the ERI phase. The ratio of crystal phases between CHA and ERI in this type of intergrowth could be varied systematically by changing both the OSDA/Si and K/Si ratios. Two intergrown zeolite samples with different Si/Al molar ratios were tested for the selective catalytic reduction (SCR) of NOx with NH3, showing competitive catalytic performance and hydrothermal stability compared to that of the industry-standard commercial NH3-SCR catalyst, Cu-SSZ-13, prevalent in automotive applications. Collectively, this work underscores the potential of our approach for the synthesis and optimization of adjustable intergrown zeolite structures, offering competitive alternatives for key industrial processes.

Autonomous, multiproperty-driven molecular discovery: From predictions to measurements and back.

A closed-loop, autonomous molecular discovery platform driven by integrated machine learning tools was developed to accelerate the design of molecules with desired properties. We demonstrated two case studies on dye-like molecules, targeting absorption wavelength, lipophilicity, and photooxidative stability. In the first study, the platform experimentally realized 294 unreported molecules across three automatic iterations of molecular design-make-test-analyze cycles while exploring the structure-function space of four rarely reported scaffolds. In each iteration, the property prediction models that guided exploration learned the structure-property space of diverse scaffold derivatives, which were realized with multistep syntheses and a variety of reactions. The second study exploited property models trained on the explored chemical space and previously reported molecules to discover nine top-performing molecules within a lightly explored structure-property space.

Examining graph neural networks for crystal structures: Limitations and opportunities for capturing periodicity.

Graph neural networks (GNNs) have recently been used to learn the representations of crystal structures through an end-to-end data-driven approach. However, a systematic top-down approach to evaluate and understand the limitations of GNNs in accurately capturing crystal structures has yet to be established. In this study, we introduce an approach using human-designed descriptors as a compendium of human knowledge to investigate the extent to which GNNs can comprehend crystal structures. Our findings reveal that current state-of-the-art GNNs fall short in accurately capturing the periodicity of crystal structures. We analyze this failure by exploring three aspects: local expressive power, long-range information processing, and readout function. To address these identified limitations, we propose a straightforward and general solution: the hybridization of descriptors with GNNs, which directly supplements the missing information to GNNs. The hybridization enhances the predictive accuracy of GNNs for specific material properties, most notably phonon internal energy and heat capacity, which heavily rely on the periodicity of materials.

Effect of Framework Composition and NH3 on the Diffusion of Cu+ in Cu-CHA Catalysts Predicted by Machine-Learning Accelerated Molecular Dynamics.

Cu-exchanged zeolites rely on mobile solvated Cu+ cations for their catalytic activity, but the role of the framework composition in transport is not fully understood. Ab initio molecular dynamics simulations can provide quantitative atomistic insight but are too computationally expensive to explore large length and time scales or diverse compositions. We report a machine-learning interatomic potential that accurately reproduces ab initio results and effectively generalizes to allow multinanosecond simulations of large supercells and diverse chemical compositions. Biased and unbiased simulations of [Cu(NH3)2]+ mobility show that aluminum pairing in eight-membered rings accelerates local hopping and demonstrate that increased NH3 concentration enhances long-range diffusion. The probability of finding two [Cu(NH3)2]+ complexes in the same cage, which is key for SCR-NOx reaction, increases with Cu content and Al content but does not correlate with the long-range mobility of Cu+. Supporting experimental evidence was obtained from reactivity tests of Cu-CHA catalysts with a controlled chemical composition.

A Model Ensemble Approach Enables Data-Driven Property Prediction for Chemically Deconstructable Thermosets in the Low-Data Regime.

Thermosets present sustainability challenges that could potentially be addressed through the design of deconstructable variants with tunable properties; however, the combinatorial space of possible thermoset molecular building blocks (e.g., monomers, cross-linkers, and additives) and manufacturing conditions is vast, and predictive knowledge for how combinations of these molecular components translate to bulk thermoset properties is lacking. Data science could overcome these problems, but computational methods are difficult to apply to multicomponent, amorphous, statistical copolymer materials for which little data exist. Here, leveraging a data set with 101 examples, we introduce a closed-loop experimental, machine learning (ML), and virtual screening strategy to enable predictions of the glass transition temperature (Tg) of polydicyclopentadiene (pDCPD) thermosets containing cleavable bifunctional silyl ether (BSE) comonomers and/or cross-linkers with varied compositions and loadings. Molecular features and formulation variables are used as model inputs, and uncertainty is quantified through model ensembling, which together with heavy regularization helps to avoid overfitting and ultimately achieves predictions within <15 °C for thermosets with compositionally diverse BSEs. This work offers a path to predicting the properties of thermosets based on their molecular building blocks, which may accelerate the discovery of promising plastics, rubbers, and composites with improved functionality and controlled deconstructability.

Design of Cytotoxic T Cell Epitopes by Machine Learning of Human Degrons.

Antigen processing is critical for producing epitope peptides that are presented by HLA molecules for T cell recognition. Therapeutic vaccines aim to harness these epitopes for priming cytotoxic T cell responses against cancer and pathogens, but insufficient processing often reduces vaccine efficacy through limiting the quantity of epitopes released. Here, we set out to improve antigen processing by harnessing protein degradation signals called degrons from the ubiquitin-proteasome system. We used machine learning to generate a computational model that ascribes a proteasomal degradation score between 0 and 100. Epitope peptides with varying degron activities were synthesized and translocated into cells using nontoxic anthrax proteins: protective antigen (PA) and the N-terminus of lethal factor (LFN). Immunogenicity studies revealed epitope sequences with a low score (<25) show pronounced T-cell activation but epitope sequences with a higher score (>75) provide limited activation. This work sheds light on the sequence-activity relationships between proteasomal degradation and epitope immunogenicity, through conserving the epitope region but varying the flanking sequence. We anticipate that future efforts to incorporate proteasomal degradation signals into vaccine designs will lead to enhanced cytotoxic T cell priming by vaccine therapeutics in clinical settings.

Mapping the Space of Photoswitchable Ligands and Photodruggable Proteins with Computational Modeling.

Light-activated drugs are a promising way to localize biological activity and minimize side effects. However, their development is complicated by the numerous photophysical and biological properties that must be simultaneously optimized. To accelerate the design of photoactive drugs, we describe a procedure that combines ligand-protein docking with chemical property prediction based on machine learning (ML). We apply this procedure to 58 proteins and 9000 photo-drug candidates based on azobenzene cis-trans isomerism. We find that most proteins display a preference for trans isomers over cis and that the binding affinities of nominally active/inactive pairs are in fact highly correlated. These findings have significant value for photopharmacology research, and reinforce the need for virtual screening to identify compounds with rare desirable properties. Further, we combine our procedure with quantum chemical validation to identify promising candidates for the photoactive inhibition of PARP1, an enzyme that is over-expressed in cancer cells. The top compounds are predicted to have long-lived active forms, differential bioactivity, and absorption in the near-infrared therapeutic window.

Entropy and Energy Profiles of Chemical Reactions.

The description of chemical processes at the molecular level is often facilitated by the use of reaction coordinates or collective variables (CVs). The CV measures the progress of the reaction and allows the construction of profiles that track how specific properties evolve as the reaction progresses. Whereas CVs are routinely used, especially alongside enhanced sampling techniques, the links among reaction profiles, thermodynamic state functions, and reaction rate constants are not rigorously exploited. Here, we report a unified treatment of such reaction profiles. Tractable expressions are derived for the free-energy, internal-energy, and entropy profiles as functions of only the CV. We demonstrate the ability of this treatment to extract quantitative insight from the entropy and internal-energy profiles of various real-world physicochemical processes, including intramolecular organic reactions, ionic transport in superionic electrolytes, and molecular transport in nanoporous materials.

Lamellar Ionenes with Highly Dissociative, Anionic Channels Provide Lower Barriers for Cation Transport.

Solid polymer electrolytes have the potential to enable safer and more energy-dense batteries; however, a deeper understanding of their ion conduction mechanisms, and how they can be optimized by molecular design, is needed to realize this goal. Here, we investigate the impact of anion dissociation energy on ion conduction in solid polymer electrolytes via a novel class of ionenes prepared using acyclic diene metathesis (ADMET) polymerization of highly dissociative, liquid crystalline fluorinated aryl sulfonimide-tagged ("FAST") anion monomers. These ionenes with various cations (Li+, Na+, K+, and Cs+) form well-ordered lamellae that are thermally stable up to 180 °C and feature domain spacings that correlate with cation size, providing channels lined with dissociative FAST anions. Electrochemical impedance spectroscopy (EIS) and differential scanning calorimetry (DSC) experiments, along with nudged elastic band (NEB) calculations, suggest that cation motion in these materials operates via an ion-hopping mechanism. The activation energy for Li+ conduction is 59 kJ/mol, which is among the lowest for systems that are proposed to operate via an ion conduction mechanism that is decoupled from polymer segmental motion. Moreover, the addition of a cation-coordinating solvent to these materials led to a >1000-fold increase in ionic conductivity without detectable disruption of the lamellar structure, suggesting selective solvation of the lamellar ion channels. This work demonstrates that molecular design can facilitate controlled formation of dissociative anionic channels that translate to significant enhancements in ion conduction in solid polymer electrolytes.

Approaching enzymatic catalysis with zeolites or how to select one reaction mechanism competing with others.

Approaching the level of molecular recognition of enzymes with solid catalysts is a challenging goal, achieved in this work for the competing transalkylation and disproportionation of diethylbenzene catalyzed by acid zeolites. The key diaryl intermediates for the two competing reactions only differ in the number of ethyl substituents in the aromatic rings, and therefore finding a selective zeolite able to recognize this subtle difference requires an accurate balance of the stabilization of reaction intermediates and transition states inside the zeolite microporous voids. In this work we present a computational methodology that, by combining a fast high-throughput screeening of all zeolite structures able to stabilize the key intermediates with a more computationally demanding mechanistic study only on the most promising candidates, guides the selection of the zeolite structures to be synthesized. The methodology presented is validated experimentally and allows to go beyond the conventional criteria of zeolite shape-selectivity.

Thermal Half-Lives of Azobenzene Derivatives: Virtual Screening Based on Intersystem Crossing Using a Machine Learning Potential.

Molecular photoswitches are the foundation of light-activated drugs. A key photoswitch is azobenzene, which exhibits trans-cis isomerism in response to light. The thermal half-life of the cis isomer is of crucial importance, since it controls the duration of the light-induced biological effect. Here we introduce a computational tool for predicting the thermal half-lives of azobenzene derivatives. Our automated approach uses a fast and accurate machine learning potential trained on quantum chemistry data. Building on well-established earlier evidence, we argue that thermal isomerization proceeds through rotation mediated by intersystem crossing, and incorporate this mechanism into our automated workflow. We use our approach to predict the thermal half-lives of 19,000 azobenzene derivatives. We explore trends and trade-offs between barriers and absorption wavelengths, and open-source our data and software to accelerate research in photopharmacology.

Chemistry-Informed Machine Learning for Polymer Electrolyte Discovery.

Solid polymer electrolytes (SPEs) have the potential to improve lithium-ion batteries by enhancing safety and enabling higher energy densities. However, SPEs suffer from significantly lower ionic conductivity than liquid and solid ceramic electrolytes, limiting their adoption in functional batteries. To facilitate more rapid discovery of high ionic conductivity SPEs, we developed a chemistry-informed machine learning model that accurately predicts ionic conductivity of SPEs. The model was trained on SPE ionic conductivity data from hundreds of experimental publications. Our chemistry-informed model encodes the Arrhenius equation, which describes temperature activated processes, into the readout layer of a state-of-the-art message passing neural network and has significantly improved accuracy over models that do not encode temperature dependence. Chemically informed readout layers are compatible with deep learning for other property prediction tasks and are especially useful where limited training data are available. Using the trained model, ionic conductivity values were predicted for several thousand candidate SPE formulations, allowing us to identify promising candidate SPEs. We also generated predictions for several different anions in poly(ethylene oxide) and poly(trimethylene carbonate), demonstrating the utility of our model in identifying descriptors for SPE ionic conductivity.

Learning pair potentials using differentiable simulations.

Learning pair interactions from experimental or simulation data is of great interest for molecular simulations. We propose a general stochastic method for learning pair interactions from data using differentiable simulations (DiffSim). DiffSim defines a loss function based on structural observables, such as the radial distribution function, through molecular dynamics (MD) simulations. The interaction potentials are then learned directly by stochastic gradient descent, using backpropagation to calculate the gradient of the structural loss metric with respect to the interaction potential through the MD simulation. This gradient-based method is flexible and can be configured to simulate and optimize multiple systems simultaneously. For example, it is possible to simultaneously learn potentials for different temperatures or for different compositions. We demonstrate the approach by recovering simple pair potentials, such as Lennard-Jones systems, from radial distribution functions. We find that DiffSim can be used to probe a wider functional space of pair potentials compared with traditional methods like iterative Boltzmann inversion. We show that our methods can be used to simultaneously fit potentials for simulations at different compositions and temperatures to improve the transferability of the learned potentials.

Machine-learning-accelerated simulations to enable automatic surface reconstruction.

Understanding material surfaces and interfaces is vital in applications such as catalysis or electronics. By combining energies from electronic structure with statistical mechanics, ab initio simulations can, in principle, predict the structure of material surfaces as a function of thermodynamic variables. However, accurate energy simulations are prohibitive when coupled to the vast phase space that must be statistically sampled. Here we present a bi-faceted computational loop to predict surface phase diagrams of multicomponent materials that accelerates both the energy scoring and statistical sampling methods. Fast, scalable and data-efficient machine learning interatomic potentials are trained on high-throughput density-functional-theory calculations through closed-loop active learning. Markov chain Monte Carlo sampling in the semigrand canonical ensemble is enabled by using virtual surface sites. The predicted surfaces for GaN(0001), Si(111) and SrTiO3(001) are in agreement with past work and indicate that the proposed strategy can model complex material surfaces and discover previously unreported surface terminations.

Expanded [23]-Helicene with Exceptional Chiroptical Properties via an Iterative Ring-Fusion Strategy.

Expanded helicenes are an emerging class of helical nanocarbons composed of alternating linear and angularly fused rings, which give rise to an internal cavity and a large diameter. The latter is expected to impart exceptional chiroptical properties, but low enantiomerization free energy barriers (ΔG‡e) have largely precluded experimental interrogation of this prediction. Here, we report the syntheses of expanded helicenes containing 15, 19, and 23 rings on the inner helical circuit, using two iterations of an Ir-catalyzed, site-selective [2 + 2 + 2] reaction. This series of compounds displays a linear relationship between the number of rings and ΔG‡e. The expanded [23]-helicene, which is 7 rings longer than any known single carbohelicene and among the longest known all-carbon ladder oligomers, exhibits a ΔG‡e that is high enough (29.2 ± 0.1 kcal/mol at 100 °C in o-DCB) to halt enantiomerization at ambient temperature. This enabled the isolation of enantiopure samples displaying circular dichroism dissymmetry factors of ±0.056 at 428 nm, which are ≥1.7× larger than values for previously reported classical and expanded helicenes. Computational investigations suggest that this improved performance is the result of both the increased diameter and length of the [23]-helicene, providing guiding design principles for high dissymmetry molecular materials.

Machine Learning Guides Peptide Nucleic Acid Flow Synthesis and Sequence Design.

Peptide nucleic acids (PNAs) are potential antisense therapies for genetic, acquired, and viral diseases. Efficiently selecting candidate PNA sequences for synthesis and evaluation from a genome containing hundreds to thousands of options can be challenging. To facilitate this process, this work leverages machine learning (ML) algorithms and automated synthesis technology to predict PNA synthesis efficiency and guide rational PNA sequence design. The training data is collected from individual fluorenylmethyloxycarbonyl (Fmoc) deprotection reactions performed on a fully automated PNA synthesizer. The optimized ML model allows for 93% prediction accuracy and 0.97 Pearson's r. The predicted synthesis scores are validated to be correlated with the experimental high-performance liquid chromatography (HPLC) crude purities (correlation coefficient R2 = 0.95). Furthermore, a general applicability of ML is demonstrated through designing synthetically accessible antisense PNA sequences from 102 315 predicted candidates targeting exon 44 of the human dystrophin gene, SARS-CoV-2, HIV, as well as selected genes associated with cardiovascular diseases, type II diabetes, and various cancers. Collectively, ML provides an accurate prediction of PNA synthesis quality and serves as a useful computational tool for informing PNA sequence design.

From free-energy profiles to activation free energies.

Given a chemical reaction going from reactant (R) to the product (P) on a potential energy surface (PES) and a collective variable (CV) discriminating between R and P, we define the free-energy profile (FEP) as the logarithm of the marginal Boltzmann distribution of the CV. This FEP is not a true free energy. Nevertheless, it is common to treat the FEP as the "free-energy" analog of the minimum potential energy path and to take the activation free energy, ΔFRP ‡, as the difference between the maximum at the transition state and the minimum at R. We show that this approximation can result in large errors. The FEP depends on the CV and is, therefore, not unique. For the same reaction, different discriminating CVs can yield different ΔFRP ‡. We derive an exact expression for the activation free energy that avoids this ambiguity. We find ΔFRP ‡ to be a combination of the probability of the system being in the reactant state, the probability density on the dividing surface, and the thermal de Broglie wavelength associated with the transition. We apply our formalism to simple analytic models and realistic chemical systems and show that the FEP-based approximation applies only at low temperatures for CVs with a small effective mass. Most chemical reactions occur on complex, high-dimensional PES that cannot be treated analytically and pose the added challenge of choosing a good CV. We study the influence of that choice and find that, while the reaction free energy is largely unaffected, ΔFRP ‡ is quite sensitive.