Unraveling Reactivity Origin of Oxygen Reduction at High-Entropy Alloy Electrocatalysts with a Computational and Data-Driven Approach.

High-entropy alloys (HEAs), characterized as compositionally complex solid solutions with five or more metal elements, have emerged as a novel class of catalytic materials with unique attributes. Because of the remarkable diversity of multielement sites or site ensembles stabilized by configurational entropy, human exploration of the multidimensional design space of HEAs presents a formidable challenge, necessitating an efficient, computational and data-driven strategy over traditional trial-and-error experimentation or physics-based modeling. Leveraging deep learning interatomic potentials for large-scale molecular simulations and pretrained machine learning models of surface reactivity, our approach effectively rationalizes the enhanced activity of a previously synthesized PdCuPtNiCo HEA nanoparticle system for electrochemical oxygen reduction, as corroborated by experimental observations. We contend that this framework deepens our fundamental understanding of the surface reactivity of high-entropy materials and fosters the accelerated development and synthesis of monodisperse HEA nanoparticles as a versatile material platform for catalyzing sustainable chemical and energy transformations.

Interpretable design of Ir-free trimetallic electrocatalysts for ammonia oxidation with graph neural networks.

The electrochemical ammonia oxidation to dinitrogen as a means for energy and environmental applications is a key technology toward the realization of a sustainable nitrogen cycle. The state-of-the-art metal catalysts including Pt and its bimetallics with Ir show promising activity, albeit suffering from high overpotentials for appreciable current densities and the soaring price of precious metals. Herein, the immense design space of ternary Pt alloy nanostructures is explored by graph neural networks trained on ab initio data for concurrently predicting site reactivity, surface stability, and catalyst synthesizability descriptors. Among a few Ir-free candidates that emerge from the active learning workflow, Pt3Ru-M (M: Fe, Co, or Ni) alloys were successfully synthesized and experimentally verified to be more active toward ammonia oxidation than Pt, Pt3Ir, and Pt3Ru. More importantly, feature attribution analyses using the machine-learned representation of site motifs provide fundamental insights into chemical bonding at metal surfaces and shed light on design strategies for high-performance catalytic systems beyond the d-band center metric of binding sites.

Novel covalent and non-covalent complex-based pharmacophore models of SARS-CoV-2 main protease (Mpro) elucidated by microsecond MD simulations.

As the world enters its second year of the pandemic caused by SARS-CoV-2, intense efforts have been directed to develop an effective diagnosis, prevention, and treatment strategies. One promising drug target to design COVID-19 treatments is the SARS-CoV-2 Mpro. To date, a comparative understanding of Mpro dynamic stereoelectronic interactions with either covalent or non-covalent inhibitors (depending on their interaction with a pocket called S1' or oxyanion hole) has not been still achieved. In this study, we seek to fill this knowledge gap using a cascade in silico protocol of docking, molecular dynamics simulations, and MM/PBSA in order to elucidate pharmacophore models for both types of inhibitors. After docking and MD analysis, a set of complex-based pharmacophore models was elucidated for covalent and non-covalent categories making use of the residue bonding point feature. The highest ranked models exhibited ROC-AUC values of 0.93 and 0.73, respectively for each category. Interestingly, we observed that the active site region of Mpro protein-ligand complex undergoes large conformational changes, especially within the S2 and S4 subsites. The results reported in this article may be helpful in virtual screening (VS) campaigns to guide the design and discovery of novel small-molecule therapeutic agents against SARS-CoV-2 Mpro protein.

Interpretable Machine Learning of Chemical Bonding at Solid Surfaces.

Understanding the nature of chemical bonding and its variation in strength across physically tunable factors is important for the development of novel catalytic materials. One way to speed up this process is to employ machine learning (ML) algorithms with online data repositories curated from high-throughput experiments or quantum-chemical simulations. Despite the reasonable predictive performance of ML models for predicting reactivity properties of solid surfaces, the ever-growing complexity of modern algorithms, e.g., deep learning, makes them black boxes with little to no explanation. In this Perspective, we discuss recent advances of interpretable ML for opening up these black boxes from the standpoints of feature engineering, algorithm development, and post hoc analysis. We underline the pivotal role of interpretability as the foundation of next-generation ML algorithms and emerging AI platforms for driving discoveries across scientific disciplines.

Infusing theory into deep learning for interpretable reactivity prediction.

Despite recent advances of data acquisition and algorithms development, machine learning (ML) faces tremendous challenges to being adopted in practical catalyst design, largely due to its limited generalizability and poor explainability. Herein, we develop a theory-infused neural network (TinNet) approach that integrates deep learning algorithms with the well-established d-band theory of chemisorption for reactivity prediction of transition-metal surfaces. With simple adsorbates (e.g., *OH, *O, and *N) at active site ensembles as representative descriptor species, we demonstrate that the TinNet is on par with purely data-driven ML methods in prediction performance while being inherently interpretable. Incorporation of scientific knowledge of physical interactions into learning from data sheds further light on the nature of chemical bonding and opens up new avenues for ML discovery of novel motifs with desired catalytic properties.

Impact of the Mode of Extraction on the Lipidomic Profile of Oils Obtained from Selected Amazonian Fruits.

Oils and fats are important raw materials in food products, animal feed, cosmetics, and pharmaceuticals among others. The market today is dominated by oils derive, d from African palm, soybean, oilseed and animal fats. Colombia's Amazon region has endemic palms such as Euterpe precatoria (açai), Oenocarpus bataua (patawa), and Mauritia flexuosa (buriti) which grow in abundance and produce a large amount of ethereal extract. However, as these oils have never been used for any economic purpose, little is known about their chemical composition or their potential as natural ingredients for the cosmetics or food industries. In order to fill this gap, we decided to characterize the lipids present in the fruits of these palms. We began by extracting the oils using mechanical and solvent-based approaches. The oils were evaluated by quantifying the quality indices and their lipidomic profiles. The main components of these profiles were triglycerides, followed by diglycerides, fatty acids, acylcarnitine, ceramides, ergosterol, lysophosphatidylcholine, phosphatidyl ethanolamine, and sphingolipids. The results suggest that solvent extraction helped increase the diglyceride concentration in the three analyzed fruits. Unsaturated lipids were predominant in all three fruits and triolein was the most abundant compound. Characterization of the oils provides important insights into the way they might behave as potential ingredients of a range of products. The sustainable use of these oils may have considerable economic potential.

A Machine Learning Strategy for Autism Screening in Toddlers.

OBJECTIVE: Autism spectrum disorder (ASD) screening can improve prognosis via early diagnosis and intervention, but lack of time and training can deter pediatric screening. The Modified Checklist for Autism in Toddlers, Revised (M-CHAT-R) is a widely used screener but requires follow-up questions and error-prone human scoring and interpretation. We consider an automated machine learning (ML) method for overcoming barriers to ASD screening, specifically using the feedforward neural network (fNN). METHODS: The fNN technique was applied using archival M-CHAT-R data of 14,995 toddlers (age 16-30 months, 46.51% male). The 20 M-CHAT-R items were inputs, and ASD diagnosis after follow-up and diagnostic evaluation (i.e., ASD or not ASD) was the output. The sample was divided into subgroups by race (i.e., white and black), sex (i.e., boys and girls), and maternal education (i.e., below and above 15 years of education completed) to examine subgroup differences. Each subgroup was evaluated for best-performing fNN models. RESULTS: For the total sample, best results yielded 99.72% correct classification using 18 items. Best results yielded 99.92% correct classification using 14 items for white toddlers and 99.79% correct classification using 18 items for black toddlers. In boys, best results yielded 99.64% correct classification using 18 items, whereas best results yielded 99.95% correct classification using 18 items in girls. For the case when maternal education is 15 years or less (i.e., associate degree and below), best results were 99.75% correct classification when using 16 items. Results were essentially the same when maternal education was 16 years or more (i.e., above associate degree); that is, 99.70% correct classification was obtained using 16 items. CONCLUSION: The ML method was comparable to the M-CHAT-R with follow-up items in accuracy of ASD diagnosis while using fewer items. Therefore, ML may be a beneficial tool in implementing automatic, efficient scoring that negates the need for labor-intensive follow-up and circumvents human error, providing an advantage over previous screening methods.

A Temperature-, pH- and Voltage-Responsive Nanogate with a Remarkably High Factor of Change in Ion Currents due to ON/OFF Switching.

In biological cells, nuclear pore complexes (NPCs) embedded in cell membranes are capable of controlling the flow of ions, for example, Na+ , K+ , and Ca2+ by responding to stimuli, for example, pH and voltage. Inspired by NPCs, researchers have been endeavoring to develop nanogates to achieve the control of ion transport, but the developed nanogates only have a low factor of change in ion currents due to ON/OFF switching. As such nanopores with high temperature and pH responsivities were developed in this work. According to the experimental results, at a voltage of 3 V, the change in ion currents due to pH change is up to a factor of 170, which is remarkably high compared to other nanogates reported. Quantum chemical (QC) calculation results show that a protonated cytosine molecule (C+ ) and an unprotonated cytosine molecule (C) form three pairs of hydrogen bonds and consequently a nucleobase pair, CC+ , leading to the binding of various strands, assembly of a strand net, and blockage of ion transport. The nanogate developed is capable of responding to temperature change. At a voltage of 3 V, the factor of change in ion currents in response to temperature variation is as high as 110. Further experiments were performed to investigate the influence of the NaCl concentrations and small opening diameters exerted on nanogate performance.

Exploration of bulk and interface behavior of gas molecules and 1-butyl-3-methylimidazolium tetrafluoroborate ionic liquid using equilibrium and nonequilibrium molecular dynamics simulation and quantum chemical calculation.

Ionic liquids (ILs) show brilliant performance in separating gas impurities, but few researchers have performed an in-depth exploration of the bulk and interface behavior of penetrants and ILs thoroughly. In this research, we have performed a study on both molecular dynamics (MD) simulation and quantum chemical (QC) calculation to explore the transport of acetylene and ethylene in the bulk and interface regions of 1-butyl-3-methylimidazolium tetrafluoroborate ([BMIM]-[BF4]). The diffusivity, solubility and permeability of gas molecules in the bulk were researched with MD simulation first. The subdiffusion behavior of gas molecules is induced by coupling between the motion of gas molecules and the ions, and the relaxation processes of the ions after the disturbance caused by gas molecules. Then, QC calculation was performed to explore the optical geometry of ions, ion pairs and complexes of ions and penetrants, and interaction potential for pairs and complexes. Finally, nonequilibrium MD simulation was performed to explore the interface structure and properties of the IL-gas system and gas molecule behavior in the interface region. The research results may be used in the design of IL separation media.

Machine-Learning-Augmented Chemisorption Model for CO2 Electroreduction Catalyst Screening.

We present a machine-learning-augmented chemisorption model that enables fast and accurate prediction of the surface reactivity of metal alloys within a broad chemical space. Specifically, we show that artificial neural networks, a family of biologically inspired learning algorithms, trained with a set of ab initio adsorption energies and electronic fingerprints of idealized bimetallic surfaces, can capture complex, nonlinear interactions of adsorbates (e.g., *CO) on multimetallics with ∼0.1 eV error, outperforming the two-level interaction model in prediction. By leveraging scaling relations between adsorption energies of similar adsorbates, we illustrate that this integrated approach greatly facilitates high-throughput catalyst screening and, as a specific case, suggests promising {100}-terminated multimetallic alloys with improved efficiency and selectivity for CO2 electrochemical reduction to C2 species. Statistical analysis of the network response to perturbations of input features underpins our fundamental understanding of chemical bonding on metal surfaces.

Diffusion-based microfluidic PCR for "one-pot" analysis of cells.

Genetic analysis starting with cell samples often requires multi-step processing including cell lysis, DNA isolation/purification, and polymerase chain reaction (PCR) based assays. When conducted on a microfluidic platform, the compatibility among various steps often demands a complicated procedure and a complex device structure. Here we present a microfluidic device that permits a "one-pot" strategy for multi-step PCR analysis starting from cells. Taking advantage of the diffusivity difference, we replace the smaller molecules in the reaction chamber by diffusion while retaining DNA molecules inside. This simple scheme effectively removes reagents from the previous step to avoid interference and thus permits multi-step processing in the same reaction chamber. Our approach shows high efficiency for PCR and potential for a wide range of genetic analysis including assays based on single cells.

A control engineering model of calcium regulation.

CONTEXT: A control engineering perspective provides a framework for representing important mechanistic details of the calcium (Ca) regulatory system efficiently. The resulting model facilitates the testing of hypotheses about mechanisms underlying the emergence of known Ca-related pathologies. OBJECTIVE: The objective of this work is to develop a comprehensive computational model that will enable quantitative understanding of plasma Ca regulation under normal and pathological conditions. DESIGN: Ca regulation is represented as an engineering control system where physiological subprocesses are mapped onto corresponding block components (sensor, controller, actuator, and process), and underlying mechanisms are represented by differential equations. The resulting model is validated with clinical observations of induced hypo- or hypercalcemia in healthy subjects, and its applicability is demonstrated by comparing model predictions of Ca-related pathologies to corresponding clinical data. RESULTS: Our model accurately predicts clinical responses to induced hypo- and hypercalcemia in healthy subjects within a framework that facilitates the representation of Ca-related pathologies in terms of control system component defects. The model also enables a deeper understanding of the emergence of pathologies and the testing of hypotheses about related features of Ca regulation-for example, why primary hyperparathyroidism and hypoparathyroidism arise from "controller defects." CONCLUSIONS: The control engineering framework provides an efficient means of organizing the subprocesses constituting Ca regulation, thereby facilitating a fundamental understanding of this complex process. The resulting validated model's predictions are consistent with clinically observed short- and long-term dynamic characteristics of the Ca regulatory system in both healthy and diseased patients. The model also enables simulation of currently infeasible clinical tests and generates predictions of physiological variables that are currently not measurable.

The new and computationally efficient MIL-SOM algorithm: potential benefits for visualization and analysis of a large-scale high-dimensional clinically acquired geographic data.

The objective of this paper is to introduce an efficient algorithm, namely, the mathematically improved learning-self organizing map (MIL-SOM) algorithm, which speeds up the self-organizing map (SOM) training process. In the proposed MIL-SOM algorithm, the weights of Kohonen's SOM are based on the proportional-integral-derivative (PID) controller. Thus, in a typical SOM learning setting, this improvement translates to faster convergence. The basic idea is primarily motivated by the urgent need to develop algorithms with the competence to converge faster and more efficiently than conventional techniques. The MIL-SOM algorithm is tested on four training geographic datasets representing biomedical and disease informatics application domains. Experimental results show that the MIL-SOM algorithm provides a competitive, better updating procedure and performance, good robustness, and it runs faster than Kohonen's SOM.

Escherichia coli autoinducer-2 uptake network does not display hysteretic behavior but AI-2 synthesis rate controls transient bifurcation.

Analysis of different architectures of quorum sensing networks has been the center of attention in recent times. The approach employs mathematical models to uncover the factors behind the dynamics. Quorum sensing networks mostly display autoregulation such as Pseudomonas aeruginosa and Vibrio cholerae. However, Escherichia coli autoinducer-2 (AI-2) synthesis does not display autoinduction (i.e. autoregulation). This and other features have raised questions about the actual function of AI-2 inside the cell. In this paper we propose a model for lsr operon regulation which explains or at least is consistent with AI-2 uptake in E. coli. The model was employed to determine the main factors that control the concentration of the signal and the uptake activation. We investigated deterministic and stochastic variants of the network model and we found no states that could lead to the typical bistability in quorum sensing systems. However, stochastic simulations predict a transient bifurcation (positively regulated by AI-2 synthesis) that could provide some advantage in adapting to new environments. LsrR inactivation was found to play a crucial role in the uptake activation compared to AI-2 synthesis, lsr transcription and AI-2 excretion. Our hypothesis is that positive regulation of the level of expression is the main factor in understanding the function of the lsr operon. This is in contrast to the conventionally held belief that the main factor is the onset of activation.

The contribution of methotrexate exposure and host factors on transcriptional variance in human liver.

Long-term administration of methotrexate (MTX) for management of chronic inflammatory diseases is associated with risk of liver damage. In this study, we examined the transcriptional profiles of livers from patients treated with MTX. The possibility that expression signatures correlate with grade of fibrosis or underlying rheumatic disease was evaluated. Twenty-seven patients taking MTX were accrued for this study. Ten non-MTX-exposed normal liver specimens were used as controls. Global mRNA expression was assayed using oligonucleotide arrays. A total of 205 genes were significantly altered in MTX-exposed livers. Six of these genes were validated by qPCR. Two genes, CLN8 and ANKH that map to chromosomal locations previously associated with rheumatoid arthritis, were found to be elevated in MTX-exposed samples. Subsequent pathway analysis indicates that MTX exposure is associated with the following key alterations: (1) upregulation of lipid biosynthetic genes, consistent with MTX-induced steatosis, (2) downregulation of proinflammatory chemokines, consistent with the anti-inflammatory effects of MTX, and (3) elevation of complement pathway gene expression. Complement 5, shown earlier to be correlated with liver fibrosis in mice, was found to be elevated (twofold) in MTX-exposed livers. In conclusion, we have found the expression of a number of genes associated with rheumatic disease and/or MTX exposure to be significantly different. Differences in complement expression provide the rationale for future correlative studies between MTX-induced liver fibrosis and C5 alleles in order to identify patients with increased risk for fibrosis.

Systematic tuning of parameters in support vector clustering.

Clustering algorithms divide a set of observations into groups so that members of the same group share common features. In most of the algorithms, tunable parameters are set arbitrarily or by trial and error, resulting in less than optimal clustering. This paper presents a global optimization strategy for the systematic and optimal selection of parameter values associated with a clustering method. In the process, a performance criterion for the optimization model is proposed and benchmarked against popular performance criteria from the literature (namely, the Silhouette coefficient, Dunn's index, and Davies-Bouldin index). The tuning strategy is illustrated using the support vector clustering (SVC) algorithm and simulated annealing. In order to reduce the computational burden, the paper also proposes an alternative to the adjacency matrix method (used for the assignment of cluster labels), namely the contour plotting approach. Datasets tested include the iris and the thyroid datasets from the UCI repository, as well as lymphoma and breast cancer data. The optimal tuning parameters are determined efficiently, while the contour plotting approach leads to significant reductions in computational effort (CPU time) especially for large datasets. The performance criteria comparisons indicate mixed results. Specifically, the Silhouette coefficient and the Davies-Bouldin index perform better, while the Dunn's index is worse on average than the proposed performance index.

Expression profile of osteoblast lineage at defined stages of differentiation.

The inherent heterogeneity of bone cells complicates the interpretation of microarray studies designed to identify genes highly associated with osteoblast differentiation. To overcome this problem, we have utilized Col1a1 promoter-green fluorescent protein transgenic mouse lines to isolate bone cells at distinct stages of osteoprogenitor maturation. Comparison of gene expression patterns from unsorted or isolated sorted bone cell populations at days 7 and 17 of calvarial cultures revealed an increased specificity regarding which genes are selectively expressed in a subset of bone cell types during differentiation. Furthermore, distinctly different patterns of gene expression associated with major signaling pathways (Igf1, Bmp, and Wnt) were observed at different levels of maturation. Some of our data differ from current models of osteoprogenitor cell differentiation and emphasize components of the pathways that were not revealed in studies based on a total cell population. Thus, applying methods to generate more homogeneous populations of cells at a defined level of cellular differentiation from a primary osteogenic culture is feasible and leads to a novel interpretation of the gene expression associated with increasing levels of osteoprogenitor maturation.

An adaptive strategy for single- and multi-cluster gene assignment.

Strict assignment of genes to one class, dimensionality reduction, a priori specification of the number of classes, the need for a training set, nonunique solution, and complex learning mechanisms are some of the inadequacies of current clustering algorithms. Existing algorithms cluster genes on the basis of high positive correlations between their expression patterns. However, genes with strong negative correlations can also have similar functions and are most likely to have a role in the same pathways. To address some of these issues, we propose the adaptive centroid algorithm (ACA), which employs an analysis of variance (ANOVA)-based performance criterion. The ACA also uses Euclidian distances, the center-of-mass principle for heterogeneously distributed mass elements, and the given data set to give unique solutions. The proposed approach involves three stages. In the first stage a two-way ANOVA of the gene expression matrix is performed. The two factors in the ANOVA are gene expression and experimental condition. The residual mean squared error (MSE) from the ANOVA is used as a performance criterion in the ACA. Finally, correlated clusters are found based on the Pearson correlation coefficients. To validate the proposed approach, a two-way ANOVA is again performed on the discovered clusters. The results from this last step indicate that MSEs of the clusters are significantly lower compared to that of the fibroblast-serum gene expression matrix. The ACA is employed in this study for single- as well as multi-cluster gene assignments.