Development of Deep Learning Models for Predicting the Effects of Exposure to Engineered Nanomaterials on Daphnia magna

This study presents the results of applying deep learning methodologies within the ecotoxicology field, with the objective of training predictive models that can support hazard assessment and eventually the design of safer engineered nanomaterials (ENMs). A workflow applying two different deep learning architectures on microscopic images of Daphnia magna is proposed that can automatically detect possible malformations, such as effects on the length of the tail, and the overall size, and uncommon lipid concentrations and lipid deposit shapes, which are due to direct or parental exposure to ENMs. Next, classification models assign specific objects (heart, abdomen/claw) to classes that depend on lipid densities and compare the results with controls. The models are statistically validated in terms of their prediction accuracy on external D. magna images and illustrate that deep learning technologies can be useful in the nanoinformatics field, because they can automate time‐consuming manual procedures, accelerate the investigation of adverse effects of ENMs, and facilitate the process of designing safer nanostructures. It may even be possible in the future to predict impacts on subsequent generations from images of parental exposure, reducing the time and cost involved in long‐term reproductive toxicity assays over multiple generations.     

Preschool predictors of narrative writing skills in elementary school children.

This study examined the preschool predictors of elementary school narrative writing skills. The sample included 65 typically developing African American children, ranging in age from 5.0 to 5.5 years, and was 44.6% male. Targeted preschool predictors included measures of phonological processing, core language abilities, prereading skills, and early writing concepts assessed during the spring or summer, just before beginning kindergarten. Using hierarchical linear modeling, findings showed that core language abilities, prereading skills, and maternal education at preschool significantly predicted the level of writing in Grades 3–5, but only core language abilities and prereading skills significantly predicted the rate of growth in writing. When kindergartners were separated into low and high readers, and low and high core language abilities, a significant pattern of widening differences emerged between the groups over time. These findings point to core language abilities, prereading skills, and maternal education assessed at kindergarten entry as critical predictors of later narrative writing skills, and they suggest the importance of including such measures when screening for written language problems in early kindergarten and early elementary school. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

A Bayesian reconstruction method for micro-rotation imaging in light microscopy

The authors present a three-dimensional (3D) reconstruction algorithm and reconstruction-based deblurring method for light microscopy using a micro-rotation device. In contrast to conventional 3D optical imaging where the focal plane is shifted along the optical axis, micro-rotation imaging employs dielectric fields to rotate the object inside a fixed optical set-up. To address this entirely new 3D-imaging modality, the authors present a reconstruction algorithm based on Bayesian inversion theory and use the total variation function as a structure prior. The spectral properties of the reconstruction by simulations that illustrate the strengths and the weaknesses of the micro-rotation approach, compared with conventional 3D optical imaging, were studied. The reconstruction from real data sets shows that this method is promising for 3D reconstruction and offers itself as a deblurring method using a reconstruction-based procedure for removing out-of-focus light from the micro-rotation image series.     

Observing How Glutathione and S-Hexyl Glutathione Bind to Glutathione S-Transferase from Rhipicephalus (Boophilus) microplus

Rhipicephalus (Boophilus) microplus is one of the most widespread ticks causing a massive loss to livestock production. The long-term use of acaracides rapidly develops acaracide resistance. In R. microplus, enhancing the metabolic activity of glutathione S-transferase (RmGST) is one of the mechanisms underlying acaracide resistance. RmGST catalyzes the conjugation of glutathione (GSH) to insecticides causing an easy-to-excrete conjugate. The active RmGST dimer contains two active sites (hydrophobic co-substrate binding site (H-site) and GSH binding site (G-site)) in each monomer. To preserve the insecticide efficacy, s-hexyl glutathione (GTX), a GST inhibitor, has been used as a synergist. To date, no molecular information on the RmGST-GSH/GTX complex is available. The insight is important for developing a novel RmGST inhibitor. Therefore, in this work, molecular dynamics simulations (MD) were performed to explore the binding of GTX and GSH to RmGST. GSH binds tighter and sits rigidly inside the G-site, while flexible GTX occupies both active sites. In GSH, the backbone mainly interacts with W8, R43, W46, K50, N59, L60, Q72, and S73, while its thiol group directs to Y7. In contrast, the aliphatic hexyl of GTX protrudes into the H-site and allows a flexible peptide core to form various interactions. Such high GTX flexibility and the protrusion of its hexyl moiety to the H-site suggest the dual role of GTX in preventing the conjugation reaction and the binding of acaracide. This insight can provide a better understanding of an important insecticide-resistance mechanism, which may in turn facilitate the development of novel approaches to tick control.     

Enhanced Ca2+ Entry Sustains the Activation of Akt in Glucose Deprived SH-SY5Y Cells

The two crucial cellular insults that take place during cerebral ischemia are the loss of oxygen and loss of glucose, which can both activate a cascade of events leading to neuronal death. In addition, the toxic overactivation of neuronal excitatory receptors, leading to Ca²⁺ overload, may contribute to ischemic neuronal injury. Brain ischemia can be simulated in vitro by oxygen/glucose deprivation, which can be reversible by the re-establishment of physiological conditions. Accordingly, we examined the effects of glucose deprivation on the PI3K/Akt survival signaling pathway and its crosstalk with HIF-1α and Ca²⁺ homeostasis in SH-SY5Y human neuroblastoma cells. It was found that glucose withdrawal decreased HIF-1α protein levels even in the presence of the ischemia-mimicking CoCl_2. On the contrary, and despite neuronal death, we identified a strong activation of the master pro-survival kinase Akt, a finding that was also confirmed by the increased phosphorylation of GSK3, a direct target of p-Akt. Remarkably, the elevated Ca²⁺ influx recorded was found to promptly trigger the activation of Akt, while a re-addition of glucose resulted in rapid restoration of both Ca²⁺ entry and p-Akt levels, highlighting the plasticity of neurons to respond to ischemic challenges and the important role of glucose homeostasis for multiple neurological disorders.     

Comparative studies between classic and wireless electrochemical promotion of a Pt catalyst for ethylene oxidation

A comparative study between a classic and a wireless electrochemical promotion experiment was undertaken as a tool towards the better understanding of both systems. The catalytic modification of a platinum catalyst for ethylene oxidation was studied. The catalyst was supported on yttria-stabilised-zirconia (YSZ), a known pure oxide ion conductor, for the classic experiment and La_0.6Sr_0.4Co_0.2Fe_0.8O_3–δ—a mixed oxide ion electronic conductor—was used for the wireless experiment. The two systems showed certain similarities in terms of the reaction classification (in both cases electrophobic behaviour was observed) and the promotion mechanism. Significant difference was observed in the time scales and the reversibility of the induced rate modification.     

Model-based recurrent neural network for modeling nonlinear dynamicsystems

A model-based recurrent neural network (MBRNN) is introduced for modeling dynamic systems. This network has a fixed structure that is defined according to the linearized state-space model of the plant. Therefore, the MBRNN has the ability to incorporate the analytical knowledge of the plant in its formulation. With its original topology intact, the MBRNN can then be trained to represent the plant nonlinearities through modifying its nodes' activation functions, which consist of contours of Gaussian radial basis functions (RBFs). Training in MBRNN involves adjusting the weights of the RBF's so as to modify the contours representing the activation functions. The performance of the MBRNN is demonstrated via several examples. The results indicate that it requires much shorter training than needed by ordinary recurrent networks. This efficiency in training is attributed to the MBRNN's fixed topology which is independent of training.     

Pattern classifier for fault diagnosis of helicopter gearboxes

Application of a diagnostic system to a helicopter gearbox is presented. The diagnostic system is a nonparametric pattern classifier that uses a multi-valued influence matrix (MVIM) as its diagnostic model and benefits from a fast learning algorithm that enables it to estimate its diagnostic model from a small number of measurement-fault data. To test this diagnostic system, vibration measurements were collected from a helicopter gearbox test stand during accelerated fatigue tests and at various fault instances. The diagnostic results indicate that the MVIM system can accurately detect and diagnose various gearbox faults so long as they are included in training.