RNA molecules cannot fold in the absence of counterions. Experiments are typically performed in the presence of monovalent and divalent cations. How to treat the impact of a solution containing a ...mixture of both ion types on RNA folding has remained a challenging problem for decades. By exploiting the large concentration difference between divalent and monovalent ions used in experiments, we develop a theory based on the reference interaction site model (RISM), which allows us to treat divalent cations explicitly while keeping the implicit screening effect due to monovalent ions. Our theory captures both the inner shell and outer shell coordination of divalent cations to phosphate groups, which we demonstrate is crucial for an accurate calculation of RNA folding thermodynamics. The RISM theory for ion–phosphate interactions when combined with simulations based on a transferable coarse-grained model allows us to predict accurately the folding of several RNA molecules in a mixture containing monovalent and divalent ions. The calculated folding free energies and ion-preferential coefficients for RNA molecules (pseudoknots, a fragment of the rRNA, and the aptamer domain of the adenine riboswitch) are in excellent agreement with experiments over a wide range of monovalent and divalent ion concentrations. Because the theory is general, it can be readily used to investigate ion and sequence effects on DNA properties.
Studies on the phase segregation of unimolecular block copolymers (BCPs) are limited by a lack of reliable, versatile methods for the synthesis of such polymers on the preparative scale. Herein, we ...describe an advancement of Iterative Exponential Growth (IEG) wherein chiral allyl-based IEG oligomers are subjected to thiol–ene reactions and converted into unimolecular BCPs. With this strategy we have synthesized uniform BCPs with molar masses up to 12.1 kDa on ∼1 g scale. BCPs composed of decane-based side chains and either triethyleneglycol- or thioglycerol-based side chains phase-segregate into hexagonal cylinder morphologies. The assembly is not driven by side-chain crystallization, but is instead the result of amorphous BCP assembly.
Fast-Convergent Federated Learning Nguyen, Hung T.; Sehwag, Vikash; Hosseinalipour, Seyyedali ...
IEEE journal on selected areas in communications,
2021-Jan., 2021-1-00, 20210101, Letnik:
39, Številka:
1
Journal Article
Recenzirano
Odprti dostop
Federated learning has emerged recently as a promising solution for distributing machine learning tasks through modern networks of mobile devices. Recent studies have obtained lower bounds on the ...expected decrease in model loss that is achieved through each round of federated learning. However, convergence generally requires a large number of communication rounds, which induces delay in model training and is costly in terms of network resources. In this paper, we propose a fast-convergent federated learning algorithm, called <inline-formula> <tex-math notation="LaTeX">\mathsf {FOLB} </tex-math></inline-formula>, which performs intelligent sampling of devices in each round of model training to optimize the expected convergence speed. We first theoretically characterize a lower bound on improvement that can be obtained in each round if devices are selected according to the expected improvement their local models will provide to the current global model. Then, we show that <inline-formula> <tex-math notation="LaTeX">\mathsf {FOLB} </tex-math></inline-formula> obtains this bound through uniform sampling by weighting device updates according to their gradient information. <inline-formula> <tex-math notation="LaTeX">\mathsf {FOLB} </tex-math></inline-formula> is able to handle both communication and computation heterogeneity of devices by adapting the aggregations according to estimates of device's capabilities of contributing to the updates. We evaluate <inline-formula> <tex-math notation="LaTeX">\mathsf {FOLB} </tex-math></inline-formula> in comparison with existing federated learning algorithms and experimentally show its improvement in trained model accuracy, convergence speed, and/or model stability across various machine learning tasks and datasets.
Ring-opening metathesis polymerization of norbornene-based (macro)monomers is a powerful approach for the synthesis of macromolecules with diverse compositions and complex architectures. ...Nevertheless, a fundamental limitation of polymers prepared by this strategy is their lack of facile degradability, limiting their utility in a range of applications. Here we describe a class of readily available bifunctional silyl ether-based cyclic olefins that copolymerize efficiently with norbornene-based (macro)monomers to provide copolymers with backbone degradability under mildly acidic aqueous conditions and degradation rates that can be tuned over several orders of magnitude, depending on the silyl ether substituents. These monomers can be used to manipulate the in vivo biodistribution and clearance rate of polyethylene glycol-based bottlebrush polymers, as well as to synthesize linear, bottlebrush and brush-arm star copolymers with degradable segments. We expect that this work will enable preparation of degradable polymers by ROMP for biomedical applications, responsive self-assembly and improved sustainability.
In this article we derive semi-analytical/numerical solutions for transport phenomena (momentum, heat and mass transfer) in a nanofluid regime adjacent to a nonlinearly porous stretching sheet by ...means of the Homotopy analysis method (
HAM
). The governing equations are reduced to a nonlinear, coupled, non-similar, ordinary differential equation system via appropriate similarity transformations. This system is solved under physically realistic boundary conditions to compute stream function, velocity, temperature and concentration function distributions. The results of the present study are compared with numerical quadrature solutions employing a shooting technique with excellent correlation. Furthermore the current
HAM
solutions demonstrate very good correlation with the non-transpiring finite element solutions of Rana and Bhargava (Commun. Nonlinear Sci. Numer. Simul. 17:212–226,
2012
). The influence of stretching parameter, transpiration (wall suction/injection) Prandtl number, Brownian motion parameter, thermophoresis parameter and Lewis number on velocity, temperature and concentration functions is illustrated graphically. Transpiration is shown to exert a substantial influence on flow characteristics. Applications of the study include industrial nanotechnological fabrication processes.
Certain bacterial pathogens are able to evade the host immune system and persist within the human host. The consequences of persistent bacterial infections potentially include increased morbidity and ...mortality from the infection itself as well as an increased risk of dissemination of disease. Eradication of persistent infections is difficult, often requiring prolonged or repeated courses of antibiotics. During persistent infections, a population or subpopulation of bacteria exists that is refractory to traditional antibiotics, possibly in a non-replicating or metabolically altered state. This review highlights the clinical significance of persistent infections and discusses different in vitro models used to investigate the altered physiology of bacteria during persistent infections. We specifically focus on recent work establishing increased protection against oxidative stress as a key element of the altered physiologic state across different in vitro models and pathogens.
Although N-heterocyclic carbenes (NHCs) have demonstrated outstanding potential for use as surface anchors, synthetic challenges have limited their application to either large planar substrates or ...very small spherical nanoparticles. The development of a strategy to graft NHCs onto non-spherical nanomaterials, such as gold nanorods, would greatly expand their utility as surface ligands. Here, we use a bidentate thiolate-NHC-gold(I) complex that is easily grafted onto commercial cetyl trimethylammonium bromide-stabilized gold nanorods through ligand exchange. On mild reduction of the resulting surface-tethered NHC-gold(I) complexes, the gold atom attached to the NHC complex is added to the surface as an adatom, thereby precluding the need for reorganization of the underlying surface lattice upon NHC binding. The resulting thiolate-NHC-stabilized gold nanorods are stable towards excess glutathione for up to six days, and under conditions with large variations in pH, high and low temperatures, high salt concentrations, or in biological media and cell culture. We also demonstrate the utility of these nanorods for in vitro photothermal therapy.
Although it is known that RNA undergoes liquid-liquid phase separation, the interplay between the molecular driving forces and the emergent features of the condensates, such as their morphologies and ...dynamic properties, is not well understood. We introduce a coarse-grained model to simulate phase separation of trinucleotide repeat RNAs, which are implicated in neurological disorders. After establishing that the simulations reproduce key experimental findings, we show that once recruited inside the liquid droplets, the monomers transition from hairpin-like structures to extended states. Interactions between the monomers in the condensates result in the formation of an intricate and dense intermolecular network, which severely restrains the fluctuations and mobilities of the RNAs inside large droplets. In the largest densely packed high-viscosity droplets, the mobility of RNA chains is best characterized by reptation, reminiscent of the dynamics in polymer melts. Our work provides a microscopic framework for understanding liquid-liquid phase separation in RNA, which is not easily discernible in current experiments.
Digital Elevation Models (DEMs) such as Advanced Spaceborne Thermal Emission and Reflection Radiometer Global Digital Elevation Models (ASTER GDEM), or Shuttle Radar Topography Mission DEM (SRTM) are ...widely used in remote areas and non-industrial countries because of their availability rather than their accuracy. Although a global DEM can be considerably enhanced using additional reference information such as higher resolution DEMs or ground truth points, improving accuracy in areas without reference data remains a challenge. This paper develops an approach to improve the accuracy of the estimated topography by combining two complementary DEMs (ASTER GDEM 1 arc-second and SRTM DEM 1 arc-second) in regions missing reference data. The combination approach is based on formulating relationships between slopes and weights in sites with reference data. Then the developed relationships are applied to sites with similar geomorphology to determine the combination weight for each DEM without using reference data. The results indicate that combined DEMs offer significant improvements of 47% and 20% in mean bias over a mountainous site, and 16% and 58% at a low-relief site when compared with the SRTM and ASTER GDEM products, respectively. DEM-derived drainages were also found to be more accurate for the combined DEMs than the near-global DEMs in areas where reference data is not available.
•A method for improving topography accuracy in areas without ground truth data•A linear combination method is adopted to combine SRTM and ASTER GDEMs.•Weights and slope relationships are developed for areas with reference data.•Slope-weight relationships are used to get weights in areas missing reference data.