Showing papers by "Rensselaer Polytechnic Institute published in 2021"

Knowledge Graphs

Abstract: In this article, we provide a comprehensive introduction to knowledge graphs, which have recently garnered significant attention from both industry and academia in scenarios that require exploiting diverse, dynamic, large-scale collections of data. After some opening remarks, we motivate and contrast various graph-based data models, as well as languages used to query and validate knowledge graphs. We explain how knowledge can be represented and extracted using a combination of deductive and inductive techniques. We conclude with high-level future research directions for knowledge graphs.

De novo protein design by deep network hallucination.

Abstract: There has been considerable recent progress in protein structure prediction using deep neural networks to predict inter-residue distances from amino acid sequences1–3. Here we investigate whether the information captured by such networks is sufficiently rich to generate new folded proteins with sequences unrelated to those of the naturally occurring proteins used in training the models. We generate random amino acid sequences, and input them into the trRosetta structure prediction network to predict starting residue–residue distance maps, which, as expected, are quite featureless. We then carry out Monte Carlo sampling in amino acid sequence space, optimizing the contrast (Kullback–Leibler divergence) between the inter-residue distance distributions predicted by the network and background distributions averaged over all proteins. Optimization from different random starting points resulted in novel proteins spanning a wide range of sequences and predicted structures. We obtained synthetic genes encoding 129 of the network-‘hallucinated’ sequences, and expressed and purified the proteins in Escherichia coli; 27 of the proteins yielded monodisperse species with circular dichroism spectra consistent with the hallucinated structures. We determined the three-dimensional structures of three of the hallucinated proteins, two by X-ray crystallography and one by NMR, and these closely matched the hallucinated models. Thus, deep networks trained to predict native protein structures from their sequences can be inverted to design new proteins, and such networks and methods should contribute alongside traditional physics-based models to the de novo design of proteins with new functions. The trRosetta neural network was used to iteratively optimise model proteins from random 100-amino-acid sequences, resulting in ‘hallucinated’ proteins, which when expressed in bacteria closely resembled the model structures.

Widespread deoxygenation of temperate lakes

Abstract: The concentration of dissolved oxygen in aquatic systems helps to regulate biodiversity1,2, nutrient biogeochemistry3, greenhouse gas emissions4, and the quality of drinking water5. The long-term declines in dissolved oxygen concentrations in coastal and ocean waters have been linked to climate warming and human activity6,7, but little is known about the changes in dissolved oxygen concentrations in lakes. Although the solubility of dissolved oxygen decreases with increasing water temperatures, long-term lake trajectories are difficult to predict. Oxygen losses in warming lakes may be amplified by enhanced decomposition and stronger thermal stratification8,9 or oxygen may increase as a result of enhanced primary production10. Here we analyse a combined total of 45,148 dissolved oxygen and temperature profiles and calculate trends for 393 temperate lakes that span 1941 to 2017. We find that a decline in dissolved oxygen is widespread in surface and deep-water habitats. The decline in surface waters is primarily associated with reduced solubility under warmer water temperatures, although dissolved oxygen in surface waters increased in a subset of highly productive warming lakes, probably owing to increasing production of phytoplankton. By contrast, the decline in deep waters is associated with stronger thermal stratification and loss of water clarity, but not with changes in gas solubility. Our results suggest that climate change and declining water clarity have altered the physical and chemical environment of lakes. Declines in dissolved oxygen in freshwater are 2.75 to 9.3 times greater than observed in the world’s oceans6,7 and could threaten essential lake ecosystem services2,3,5,11. Analysis of temperate lakes finds a widespread decline in dissolved oxygen concentrations in surface and deep waters, which is associated with reduced solubility at warmer surface water temperatures and increased stratification at depth.

Effective Inhibition of SARS-CoV-2 Entry by Heparin and Enoxaparin Derivatives.

Abstract: Severe acute respiratory syndrome-related coronavirus 2 (SARS-CoV-2) has caused a pandemic of historic proportions and continues to spread globally, with enormous consequences to human health. Currently there is no vaccine, effective therapeutic, or prophylactic. As with other betacoronaviruses, attachment and entry of SARS-CoV-2 are mediated by the spike glycoprotein (SGP). In addition to its well-documented interaction with its receptor, human angiotensin-converting enzyme 2 (hACE2), SGP has been found to bind to glycosaminoglycans like heparan sulfate, which is found on the surface of virtually all mammalian cells. Here, we pseudotyped SARS-CoV-2 SGP on a third-generation lentiviral (pLV) vector and tested the impact of various sulfated polysaccharides on transduction efficiency in mammalian cells. The pLV vector pseudotyped SGP efficiently and produced high titers on HEK293T cells. Various sulfated polysaccharides potently neutralized pLV-S pseudotyped virus with clear structure-based differences in antiviral activity and affinity to SGP. Concentration-response curves showed that pLV-S particles were efficiently neutralized by a range of concentrations of unfractionated heparin (UFH), enoxaparin, 6-O-desulfated UFH, and 6-O-desulfated enoxaparin with 50% inhibitory concentrations (IC50s) of 5.99 µg/liter, 1.08 mg/liter, 1.77 µg/liter, and 5.86 mg/liter, respectively. In summary, several sulfated polysaccharides show potent anti-SARS-CoV-2 activity and can be developed for prophylactic as well as therapeutic purposes.IMPORTANCE The emergence of severe acute respiratory syndrome coronavirus (SARS-CoV-2) in Wuhan, China, in late 2019 and its subsequent spread to the rest of the world has created a pandemic situation unprecedented in modern history. While ACE2 has been identified as the viral receptor, cellular polysaccharides have also been implicated in virus entry. The SARS-CoV-2 spike glycoprotein (SGP) binds to glycosaminoglycans like heparan sulfate, which is found on the surface of virtually all mammalian cells. Here, we report structure-based differences in antiviral activity and affinity to SGP for several sulfated polysaccharides, including both well-characterized FDA-approved drugs and novel marine sulfated polysaccharides, which can be developed for prophylactic as well as therapeutic purposes.

Multiscale modeling meets machine learning: What can we learn?

Abstract: Machine learning is increasingly recognized as a promising technology in the biological, biomedical, and behavioral sciences. There can be no argument that this technique is incredibly successful in image recognition with immediate applications in diagnostics including electrophysiology, radiology, or pathology, where we have access to massive amounts of annotated data. However, machine learning often performs poorly in prognosis, especially when dealing with sparse data. This is a field where classical physics-based simulation seems to remain irreplaceable. In this review, we identify areas in the biomedical sciences where machine learning and multiscale modeling can mutually benefit from one another: Machine learning can integrate physics-based knowledge in the form of governing equations, boundary conditions, or constraints to manage ill-posted problems and robustly handle sparse and noisy data; multiscale modeling can integrate machine learning to create surrogate models, identify system dynamics and parameters, analyze sensitivities, and quantify uncertainty to bridge the scales and understand the emergence of function. With a view towards applications in the life sciences, we discuss the state of the art of combining machine learning and multiscale modeling, identify applications and opportunities, raise open questions, and address potential challenges and limitations. We anticipate that it will stimulate discussion within the community of computational mechanics and reach out to other disciplines including mathematics, statistics, computer science, artificial intelligence, biomedicine, systems biology, and precision medicine to join forces towards creating robust and efficient models for biological systems.

Correlated insulating states at fractional fillings of the WS2/WSe2 moiré lattice

Abstract: The strong electron interactions in the minibands formed in moire superlattices of van der Waals materials, such as twisted graphene and transition metal dichalcogenides, make such systems a fascinating platform with which to study strongly correlated states1–19. In most systems, the correlated states appear when the moire lattice is filled by an integer number of electrons per moire unit cell. Recently, correlated states at fractional fillings of 1/3 and 2/3 holes per moire unit cell have been reported in the WS2/WSe2 hetero-bilayer, hinting at the long-range nature of the electron interaction16. Here we observe a series of correlated insulating states at fractional fillings of the moire minibands on both electron- and hole-doped sides in angle-aligned WS2/WSe2 hetero-bilayers, with certain states persisting at temperatures up to 120 K. Simulations reveal that these insulating states correspond to ordering of electrons in the moire lattice with a periodicity much larger than the moire unit cell, indicating a surprisingly strong and long-range interaction beyond the nearest neighbours. Twisted bilayers of WS2 and WSe2 have correlated states that correspond to real-space ordering of the electrons on a length scale much longer than the moire pattern.

Designing anion exchange membranes for CO2 electrolysers

Abstract: New technologies are required to electrocatalytically convert carbon dioxide (CO2) into fuels and chemicals at near-ambient temperatures and pressures more effectively. One particular challenge is mediating the electrochemical CO2 reduction reaction (CO2RR) at low cell voltages while maintaining high conversion efficiencies. Anion exchange membranes (AEMs) in zero-gap reactors offer promise in this direction; however, there remain substantial obstacles to be overcome in tailoring the membranes and other cell components to the requirements of CO2RR systems. Here we review recent advances, and remaining challenges, in AEM materials and devices for CO2RR. We discuss the principles underpinning AEM operation and the properties desired for CO2RR, in addition to reviewing state-of-the-art AEMs in CO2 electrolysers. We close with future design strategies to minimize product crossover, improve mechanical and chemical stability, and overcome the energy losses associated with the use of AEMs for CO2RR systems. Carbon dioxide electroreduction is a promising approach to synthesize chemicals and fuels using renewable energy. This Review explores our understanding of anion exchange membranes — a key component of certain carbon dioxide electrolysers — and outlines approaches to design improved materials.

Durability of anion exchange membrane water electrolyzers

Abstract: Interest in the low-cost production of clean hydrogen is growing. Anion exchange membrane water electrolyzers (AEMWEs) are considered one of the most promising sustainable hydrogen production technologies because of their ability to split water using platinum group metal-free catalysts, less expensive anode flow fields, and bipolar plates. Critical to the realization of AEMWEs is understanding the durability-limiting factors that restrict the long-term use of these devices. This article presents both durability-limiting factors and mitigation strategies for AEMWEs under three operation modes, i.e., pure water-fed (no liquid electrolyte), concentrated KOH-fed, and 1 wt% K2CO3-fed operating at a differential pressure of 100 psi. We examine extended-term behaviors of AEMWEs at the single-cell level and connect their behavior with the electrochemical, chemical, and mechanical instability of single-cell components. Finally, we discuss the pros and cons of AEMWEs under these operation modes and provide direction for long-lasting AEMWEs with highly efficient hydrogen production capabilities.

Lower complexity bounds of first-order methods for convex-concave bilinear saddle-point problems

Abstract: On solving a convex-concave bilinear saddle-point problem (SPP), there have been many works studying the complexity results of first-order methods. These results are all about upper complexity bounds, which can determine at most how many iterations would guarantee a solution of desired accuracy. In this paper, we pursue the opposite direction by deriving lower complexity bounds of first-order methods on large-scale SPPs. Our results apply to the methods whose iterates are in the linear span of past first-order information, as well as more general methods that produce their iterates in an arbitrary manner based on first-order information. We first work on the affinely constrained smooth convex optimization that is a special case of SPP. Different from gradient method on unconstrained problems, we show that first-order methods on affinely constrained problems generally cannot be accelerated from the known convergence rate O(1 / t) to $$O(1/t^2)$$ , and in addition, O(1 / t) is optimal for convex problems. Moreover, we prove that for strongly convex problems, $$O(1/t^2)$$ is the best possible convergence rate, while it is known that gradient methods can have linear convergence on unconstrained problems. Then we extend these results to general SPPs. It turns out that our lower complexity bounds match with several established upper complexity bounds in the literature, and thus they are tight and indicate the optimality of several existing first-order methods.

On Interpretability of Artificial Neural Networks: A Survey

Abstract: Deep learning as performed by artificial deep neural networks (DNNs) has achieved great successes recently in many important areas that deal with text, images, videos, graphs, and so on. However, the black-box nature of DNNs has become one of the primary obstacles for their wide adoption in mission-critical applications such as medical diagnosis and therapy. Because of the huge potentials of deep learning, the interpretability of DNNs has recently attracted much research attention. In this article, we propose a simple but comprehensive taxonomy for interpretability, systematically review recent studies on interpretability of neural networks, describe applications of interpretability in medicine, and discuss future research directions, such as in relation to fuzzy logic and brain science.

Co and Pt Dual-Single-Atoms with Oxygen-Coordinated Co-O-Pt Dimer Sites for Ultrahigh Photocatalytic Hydrogen Evolution Efficiency.

Abstract: The platinum single-atom-catalyst is verified as a very successful route to approach the size limit of Pt catalysts, while how to further improve the catalytic efficiency of Pt is a fundamental scientific question and is challenging because the size issue of Pt is approached at the ultimate ceiling as single atoms. Here, a new route for further improving Pt catalytic efficiency by cobalt (Co) and Pt dual-single-atoms on titanium dioxide (TiO2 ) surfaces, which contains a fraction of nonbonding oxygen-coordinated Co-O-Pt dimers, is reported. These Co-Pt dimer sites originate from loading high-density Pt single-atoms and Co single-atoms, with them anchoring randomly on the TiO2 substrate. This dual-single-atom catalyst yields 13.4% dimer sites and exhibits an ultrahigh and stable photocatalytic activity with a rate of 43.467 mmol g-1 h-1 and external quantum efficiency of ≈83.4% at 365 nm. This activity far exceeds those of equal amounts of Pt single-atom and typical Pt clustered catalysts by 1.92 and 1.64 times, respectively. The enhancement mechanism relies on the oxygen-coordinated Co-O-Pt dimer coupling, which can mutually optimize the electronic states of both Pt and Co sites to weaken H* binding. Namely, the "mute" Co single-atom is activated by Pt single-atom and the activity of the Pt atom is further enhanced through the dimer interaction. This strategy of nonbonding interactive dimer sites and the oxygen-mediated catalytic mechanisms provide emerging rich opportunities for greatly improving the catalytic efficiency and developing novel catalysts with creating new electronic states.

A comprehensive review of DC fault protection methods in HVDC transmission systems

Abstract: High voltage direct current (HVDC) transmission is an economical option for transmitting a large amount of power over long distances. Initially, HVDC was developed using thyristor-based current source converters (CSC). With the development of semiconductor devices, a voltage source converter (VSC)-based HVDC system was introduced, and has been widely applied to integrate large-scale renewables and network interconnection. However, the VSC-based HVDC system is vulnerable to DC faults and its protection becomes ever more important with the fast growth in number of installations. In this paper, detailed characteristics of DC faults in the VSC-HVDC system are presented. The DC fault current has a large peak and steady values within a few milliseconds and thus high-speed fault detection and isolation methods are required in an HVDC grid. Therefore, development of the protection scheme for a multi-terminal VSC-based HVDC system is challenging. Various methods have been developed and this paper presents a comprehensive review of the different techniques for DC fault detection, location and isolation in both CSC and VSC-based HVDC transmission systems in two-terminal and multi-terminal network configurations.

DRONE: Dual-Domain Residual-based Optimization NEtwork for Sparse-View CT Reconstruction

Abstract: Deep learning has attracted rapidly increasing attention in the field of tomographic image reconstruction, especially for CT, MRI, PET/SPECT, ultrasound and optical imaging. Among various topics, sparse-view CT remains a challenge which targets a decent image reconstruction from very few projections. To address this challenge, in this article we propose a Dual-domain Residual-based Optimization NEtwork (DRONE). DRONE consists of three modules respectively for embedding, refinement, and awareness. In the embedding module, a sparse sinogram is first extended. Then, sparse-view artifacts are effectively suppressed in the image domain. After that, the refinement module recovers image details in the residual data and image domains synergistically. Finally, the results from the embedding and refinement modules in the data and image domains are regularized for optimized image quality in the awareness module, which ensures the consistency between measurements and images with the kernel awareness of compressed sensing. The DRONE network is trained, validated, and tested on preclinical and clinical datasets, demonstrating its merits in edge preservation, feature recovery, and reconstruction accuracy.

Flexo-photovoltaic effect in MoS 2 .

Abstract: The theoretical Shockley-Queisser limit of photon-electricity conversion in a conventional p-n junction could be potentially overcome by the bulk photovoltaic effect that uniquely occurs in non-centrosymmetric materials. Using strain-gradient engineering, the flexo-photovoltaic effect, that is, the strain-gradient-induced bulk photovoltaic effect, can be activated in centrosymmetric semiconductors, considerably expanding material choices for future sensing and energy applications. Here we report an experimental demonstration of the flexo-photovoltaic effect in an archetypal two-dimensional material, MoS2, by using a strain-gradient engineering approach based on the structural inhomogeneity and phase transition of a hybrid system consisting of MoS2 and VO2. The experimental bulk photovoltaic coefficient in MoS2 is orders of magnitude higher than that in most non-centrosymmetric materials. Our findings unveil the fundamental relation between the flexo-photovoltaic effect and a strain gradient in low-dimensional materials, which could potentially inspire the exploration of new optoelectronic phenomena in strain-gradient-engineered materials.

Evaluation of individual and ensemble probabilistic forecasts of COVID-19 mortality in the US

Abstract: Short-term probabilistic forecasts of the trajectory of the COVID-19 pandemic in the United States have served as a visible and important communication channel between the scientific modeling community and both the general public and decision-makers Forecasting models provide specific, quantitative, and evaluable predictions that inform short-term decisions such as healthcare staffing needs, school closures, and allocation of medical supplies In 2020, the COVID-19 Forecast Hub (https://covid19forecasthuborg/) collected, disseminated, and synthesized hundreds of thousands of specific predictions from more than 50 different academic, industry, and independent research groups This manuscript systematically evaluates 23 models that regularly submitted forecasts of reported weekly incident COVID-19 mortality counts in the US at the state and national level One of these models was a multi-model ensemble that combined all available forecasts each week The performance of individual models showed high variability across time, geospatial units, and forecast horizons Half of the models evaluated showed better accuracy than a naive baseline model In combining the forecasts from all teams, the ensemble showed the best overall probabilistic accuracy of any model Forecast accuracy degraded as models made predictions farther into the future, with probabilistic accuracy at a 20-week horizon more than 5 times worse than when predicting at a 1-week horizon This project underscores the role that collaboration and active coordination between governmental public health agencies, academic modeling teams, and industry partners can play in developing modern modeling capabilities to support local, state, and federal response to outbreaks

EGCG binds intrinsically disordered N-terminal domain of p53 and disrupts p53-MDM2 interaction

Abstract: Epigallocatechin gallate (EGCG) from green tea can induce apoptosis in cancerous cells, but the underlying molecular mechanisms remain poorly understood Using SPR and NMR, here we report a direct, μM interaction between EGCG and the tumor suppressor p53 (KD = 16 ± 14 μM), with the disordered N-terminal domain (NTD) identified as the major binding site (KD = 4 ± 2 μM) Large scale atomistic simulations (>100 μs), SAXS and AUC demonstrate that EGCG-NTD interaction is dynamic and EGCG causes the emergence of a subpopulation of compact bound conformations The EGCG-p53 interaction disrupts p53 interaction with its regulatory E3 ligase MDM2 and inhibits ubiquitination of p53 by MDM2 in an in vitro ubiquitination assay, likely stabilizing p53 for anti-tumor activity Our work provides insights into the mechanisms for EGCG’s anticancer activity and identifies p53 NTD as a target for cancer drug discovery through dynamic interactions with small molecules Epigallocatechin gallate (EGCG) is a catechin flavonoid which induces apoptosis in cancerous cells, but the underlying molecular mechanisms remain poorly understood Here authors use an interdisciplinary approach to show a direct interaction between EGCG and the tumor suppressor p53 and demonstrate that EGCG inhibits ubiquitination of p53 by MDM2

Solving Stochastic Compositional Optimization is Nearly as Easy as Solving Stochastic Optimization

Abstract: Stochastic compositional optimization generalizes classic (non-compositional) stochastic optimization to the minimization of compositions of functions. Each composition may introduce an additional expectation. The series of expectations may be nested. Stochastic compositional optimization is gaining popularity in applications such as reinforcement learning and meta learning. This paper presents a new S tochastically C orrected S tochastic C ompositional gradient method ( SCSC ). SCSC runs in a single-time scale with a single loop, uses a fixed batch size, and guarantees to converge at the same rate as the stochastic gradient descent (SGD) method for non-compositional stochastic optimization. This is achieved by making a careful improvement to a popular stochastic compositional gradient method. It is easy to apply SGD-improvement techniques to accelerate SCSC. This helps SCSC achieve state-of-the-art performance for stochastic compositional optimization. In particular, we apply Adam to SCSC, and the exhibited rate of convergence matches that of the original Adam on non-compositional stochastic optimization. We test SCSC using the model-agnostic meta-learning tasks.

Search for Coherent Elastic Scattering of Solar ^{8}B Neutrinos in the XENON1T Dark Matter Experiment

Abstract: We report on a search for nuclear recoil signals from solar $^8$B neutrinos elastically scattering off xenon nuclei in XENON1T data, lowering the energy threshold from 2.6 keV to 1.6 keV. We develop a variety of novel techniques to limit the resulting increase in backgrounds near the threshold. No significant $^8$B neutrino-like excess is found in an exposure of 0.6 t $\times$ y. For the first time, we use the non-detection of solar neutrinos to constrain the light yield from 1-2 keV nuclear recoils in liquid xenon, as well as non-standard neutrino-quark interactions. Finally, we improve upon world-leading constraints on dark matter-nucleus interactions for dark matter masses between 3 GeV/c$^2$ and 11 GeV/c$^2$ by as much as an order of magnitude.

Carbon Dots: A Future Blood–Brain Barrier Penetrating Nanomedicine and Drug Nanocarrier

Abstract: Drug delivery across the blood-brain barrier (BBB) is one of the biggest challenges in modern medicine due to the BBB's highly semipermeable property that limits most therapeutic agents of brain diseases to enter the central nervous system (CNS). In recent years, nanoparticles, especially carbon dots (CDs), exhibit many unprecedented applications for drug delivery. Several types of CDs and CD-ligand conjugates have been reported successfully penetrating the BBB, which shows a promising progress in the application of CD-based drug delivery system (DDS) for the treatment of CNS diseases. In this review, our discussion of CDs includes their classification, preparations, structures, properties, and applications for the treatment of neurodegenerative diseases, especially Alzheimer's disease (AD) and brain tumor. Moreover, abundant functional groups on the surface, especially amine and carboxyl groups, allow CDs to conjugate with diverse drugs as versatile drug nanocarriers. In addition, structure of the BBB is briefly described, and mechanisms for transporting various molecules across the BBB and other biological barriers are elucidated. Most importantly, recent developments in drug delivery with CDs as BBB-penetrating nanodrugs and drug nanocarriers to target CNS diseases especially Alzheimer's disease and brain tumor are summarized. Eventually, future prospects of the CD-based DDS are discussed in combination with the development of artificial intelligence and nanorobots.

Investigation of Mechanical, Material, and Compositional Determinants of Human Trabecular Bone Quality in Type 2 Diabetes

Abstract: Context Increased bone fragility and reduced energy absorption to fracture associated with type 2 diabetes (T2D) cannot be explained by bone mineral density alone. This study, for the first time, reports on alterations in bone tissue's material properties obtained from individuals with diabetes and known fragility fracture status. Objective To investigate the role of T2D in altering biomechanical, microstructural, and compositional properties of bone in individuals with fragility fracture. Methods Femoral head bone tissue specimens were collected from patients who underwent replacement surgery for fragility hip fracture. Trabecular bone quality parameters were compared in samples of 2 groups, nondiabetic (n = 40) and diabetic (n = 30), with a mean duration of disease 7.5 ± 2.8 years. Results No significant difference was observed in aBMD between the groups. Bone volume fraction (BV/TV) was lower in the diabetic group due to fewer and thinner trabeculae. The apparent-level toughness and postyield energy were lower in those with diabetes. Tissue-level (nanoindentation) modulus and hardness were lower in this group. Compositional differences in the diabetic group included lower mineral:matrix, wider mineral crystals, and bone collagen modifications-higher total fluorescent advanced glycation end-products (fAGEs), higher nonenzymatic cross-link ratio (NE-xLR), and altered secondary structure (amide bands). There was a strong inverse correlation between NE-xLR and postyield strain, fAGEs and postyield energy, and fAGEs and toughness. Conclusion The current study is novel in examining bone tissue in T2D following first hip fragility fracture. Our findings provide evidence of hyperglycemia's detrimental effects on trabecular bone quality at multiple scales leading to lower energy absorption and toughness indicative of increased propensity to bone fragility.