Simultaneous production of furfural, lignin and cellulose-rich residue from Eucalyptus urophylla × E. grandis by ChCl/1,2-propanediol/MIBK biphasic system pretreatment.

A facile biphasic system composed of choline chloride (ChCl)-based deep eutectic solvent (DES) and methyl isobutyl ketone (MIBK) was developed to realize the furfural production, lignin separation and preparation of fermentable glucose from Eucalyptus in one-pot. Results showed that the ChCl/1,2-propanediol/MIBK system owned the best property to convert hemicelluloses into furfural. Under the optimal conditions (MRChCl:1,2-propanediol = 1:2, raw materials:DES:MIBK ratio = 1:4:8 g/g/mL, 0.075 mol/L AlCl3·6H2O, 140 °C, and 90 min), the furfural yield and glucose yield reached 65.0 and 92.2 %, respectively. Meanwhile, the lignin with low molecular weight (1250-1930 g/mol), low polydispersity (DM = 1.25-1.53) and high purity (only 0.08-2.59 % carbohydrate content) was regenerated from the biphasic system. With the increase of pretreatment temperature, the ß-O-4, ß-ß and ß-5 linkages in the regenerated lignin were gradually broken, and the content of phenolic hydroxyl groups increased, but the content of aliphatic hydroxyl groups decreased. This research provides a new strategy for the comprehensive utilization of lignocellulose in biorefinery process.

Harnessing Renewable Lignocellulosic Potential for Sustainable Wastewater Purification.

Utilizing renewable lignocellulosic resources for wastewater remediation is crucial to achieving sustainable social development. However, the resulting by-products and the synthetic process characterized by complexity, high cost, and environmental pollution limit the further development of lignocellulose-based materials. Here, we developed a sustainable strategy that involved a new functional deep eutectic solvent (DES) to deconstruct industrial xylose residue into cellulose-rich residue with carboxyl groups, lignin with carboxyl and quaternary ammonium salt groups, and DES effluent rich in lignin fragments. Subsequently, these fractions equipped with customized functionality were used to produce efficient wastewater remediation materials in cost-effective and environmentally sound manners, namely, photocatalyst prepared by carboxyl-modified cellulose residue, biochar-based adsorbent originated from modified lignin, and flocculant synthesized by self-catalytic in situ copolymerization of residual DES effluent at room temperature. Under the no-waste principle, this strategy upgraded the whole components of waste lignocellulose into high-value-added wastewater remediation materials with excellent universality. These materials in coordination with each other can stepwise purify high-hazardous mineral processing wastewater into drinkable water, including the removal of 99.81% of suspended solids, almost all various heavy metal ions, and 97.09% chemical oxygen demand, respectively. This work provided promising solutions and blueprints for lignocellulosic resources to alleviate water shortages while also advancing the global goal of carbon neutrality.

Structure-guided engineering of branched-chain α-keto acid decarboxylase for improved 1,2,4-butanetriol production by in vitro synthetic enzymatic biosystem.

Efficient synthetic routes for biomanufacturing chemicals often require the overcoming of pathway bottlenecks by tailoring enzymes to improve the catalytic efficiency or even implement non-native activities. 1,2,4-butanetriol (BTO), a valuable commodity chemical, is currently biosynthesized from D-xylose via a four-enzyme reaction cascade, with the ThDP-dependent α-keto acid decarboxylase (KdcA) identified as the potential bottleneck. Here, to further enhance the catalytic activity of KdcA toward the non-native substrate α-keto-3-deoxy-xylonate (KDX), in silico screening and structure-guided evolution were performed. The best mutants, S286L/G402P and V461K, exhibited a 1.8- and 2.5-fold higher enzymatic activity in the conversion of KDX to 3,4-dihydroxybutanal when compared to KdcA, respectively. MD simulations revealed that the two sets of mutations reshaped the substrate binding pocket, thereby increasing the binding affinity for KDX and promoting interactions between KDX and cofactor ThDP. Then, when the V461K mutant instead of wild type KdcA was integrated into the enzyme cascade, a 1.9-fold increase in BTO titer was observed. After optimization of the reaction conditions, the enzyme cocktail contained V461K converted 60 g/L D-xylose to 22.1 g/L BTO with a yield of 52.1 %. This work illustrated that protein engineering is a powerful tool for modifying the output of metabolic pathway.

One-step conversion of corn stalk to glucose and furfural in molten salt hydrate/organic solvent biphasic system.

An effective approach for glucose and furfural production by converting cellulose and hemicelluloses from corn stalk in a biphasic system of molten salt hydrate (MSH) and organic solvent using H2SO4 as catalyst was reported. Results showed that the system with LiBr·3H2O and dichloromethane (DCM) had excellent performance in cellulose and hemicelluloses conversion. Under the optimal reaction conditions (corn stalk:LiBr·3H2O:DCM ratio = 0.35:10:20 g/mL/mL, 0.05 mol/L H2SO4, 120 °C, 90 min), 58.9% glucose and 72.5% furfural were yielded. Meanwhile, lignin was obviously depolymerized by the cleavage of ß-O-4' linkages and fractionated with high purity and low molecular weight for potential coproducts. Fluorescence microscopy and confocal Raman microscope displayed that the LiBr·3H2O/DCM treatment caused decreasing intensities in carbohydrate and lignin, suggesting the degradation of the main components of biomass. This research provided a promising biorefinery technology for the comprehensive utilization of corn stalk.

Key Challenges and Recommendations for In Vitro Testing of Tobacco Products for Regulatory Applications: Consideration of Test Materials and Exposure Parameters.

The Institute for In Vitro Sciences (IIVS) is sponsoring a series of workshops to identify, discuss and develop recommendations for optimal scientific and technical approaches for conducting in vitro assays, to assess potential toxicity within and across tobacco and various next generation nicotine and tobacco products (NGPs), including heated tobacco products (HTPs) and electronic nicotine delivery systems (ENDS). The third workshop (24-26 February 2020) summarised the key challenges and made recommendations concerning appropriate methods of test article generation and cell exposure from combustible cigarettes, HTPs and ENDS. Expert speakers provided their research, perspectives and recommendations for the three basic types of tobacco-related test articles: i) pad-collected material (PCM); ii) gas vapour phase (GVP); and iii) whole smoke/aerosol. These three types of samples can be tested individually, or the PCM and GVP can be combined. Whole smoke/aerosol can be bubbled through media or applied directly to cells at the air-liquid interface. Summaries of the speaker presentations and the recommendations developed by the workgroup are presented. Following discussion, the workshop concluded the following: that there needs to be greater standardisation in aerosol generation and collection processes; that methods for testing the NGPs need to be developed and/or optimised, since simply mirroring cigarette smoke testing approaches may be insufficient; that understanding and quantitating the applied dose is fundamental to the interpretation of data and conclusions from each study; and that whole smoke/aerosol approaches must be contextualised with regard to key information, including appropriate experimental controls, environmental conditioning, analytical monitoring, verification and performance criteria.

A self-attention integrated spatiotemporal LSTM approach to edge-radar echo extrapolation in the Internet of Radars.

In recent years, the number of weather-related disasters significantly increases across the world. As a typical example, short-range extreme precipitation can cause severe flooding and other secondary disasters, which therefore requires accurate prediction of extent and intensity of precipitation in a relatively short period of time. Based on the echo extrapolation of networked weather radars (i.e., the Internet of Radars), different solutions have been presented ranging from traditional optical-flow methods to recent deep neural networks. However, these existing networks focus on local features of echo variations to model the dynamics of holistic radar echo motion, so it often suffers from inaccurate extrapolation of the radar echo motion trend, trajectory, and intensity. To address the problem, this paper introduces the self-attention mechanism and an extra memory that saves global spatiotemporal feature into the original Spatiotemporal LSTM (ST-LSTM) to form a self-attention Integrated ST-LSTM recurrent unit (SAST-LSTM), capturing both spatial and temporal global features of radar echo motion. And several these units are stacked to build the radar echo extrapolation network SAST-Net. Comparative experiments show that the proposed model has better performance on different real world radar echo datasets over other recent methods.

Microwave-assisted choline chloride/1,2-propanediol/methyl isobutyl ketone biphasic system for one-pot fractionation and valorization of Eucalyptus biomass.

The developing of pretreatment method to break the biomass barrier of lignocellulosic is a challenging task for achieve high value utilization. A fast microwave-assisted choline chloride/1,2-propanediol/methyl isobutyl ketone biphasic system was constructed for pretreating Eucalyptus to the production of furfural and cellulose-rich residues and the extraction of lignin. Results showed that the combination of AlCl3·6H2O and HCl had the best catalytic ability for furfural production among the examined catalysts. Under the optimal conditions (140 °C, 15 min, 0.075 M AlCl3·6H2O, 0.05 M HCl), the furfural yield of 55.4 %, the glucose yield of 90.3 % and the delignification rate of 92.4 % could be achieved. Moreover, the extracted lignin samples with a low polydispersity (1.55-1.73) and molecular weight (1380-2040 g/mol) are promising to act as precursor for the value-add products processing. These findings demonstrated an ultrafast pretreatment process with excellent results in biomass fractionation and comprehensive utilization of biomass components.

Quantitative in vitro to in vivo extrapolation of genotoxicity data provides protective estimates of in vivo dose.

Genotoxicity assessment is a critical component in the development and evaluation of chemicals. Traditional genotoxicity assays (i.e., mutagenicity, clastogenicity, and aneugenicity) have been limited to dichotomous hazard classification, while other toxicity endpoints are assessed through quantitative determination of points-of-departures (PODs) for setting exposure limits. The more recent higher-throughput in vitro genotoxicity assays, many of which also provide mechanistic information, offer a powerful approach for determining defined PODs for potency ranking and risk assessment. In order to obtain relevant human dose context from the in vitro assays, in vitro to in vivo extrapolation (IVIVE) models are required to determine what dose would elicit a concentration in the body demonstrated to be genotoxic using in vitro assays. Previous work has demonstrated that application of IVIVE models to in vitro bioactivity data can provide PODs that are protective of human health, but there has been no evaluation of how these models perform with in vitro genotoxicity data. Thus, the Genetic Toxicology Technical Committee, under the Health and Environmental Sciences Institute, conducted a case study on 31 reference chemicals to evaluate the performance of IVIVE application to genotoxicity data. The results demonstrate that for most chemicals considered here (20/31), the PODs derived from in vitro data and IVIVE are health protective relative to in vivo PODs from animal studies. PODs were also protective by assay target: mutations (8/13 chemicals), micronuclei (9/12), and aneugenicity markers (4/4). It is envisioned that this novel testing strategy could enhance prioritization, rapid screening, and risk assessment of genotoxic chemicals.

Subacute Pulmonary Toxicity of Glutaraldehyde Aerosols in a Human In Vitro Airway Tissue Model.

Glutaraldehyde (GA) has been cleared by the Center for Devices and Radiological Health (CDRH) of the Food and Drug Administration (FDA) as a high-level disinfectant for disinfecting heat-sensitive medical equipment in hospitals and healthcare facilities. Inhalation exposure to GA is known to cause respiratory irritation and sensitization in animals and humans. To reproduce some of the known in vivo effects elicited by GA, we used a liquid aerosol exposure system and evaluated the tissue responses in a human in vitro airway epithelial tissue model. The cultures were treated at the air interface with various concentrations of GA aerosols on five consecutive days and changes in tissue function and structure were evaluated at select timepoints during the treatment phase and after a 7-day recovery period. Exposure to GA aerosols caused oxidative stress, inhibition of ciliary beating frequency, aberrant mucin production, and disturbance of cytokine and matrix metalloproteinase secretion, as well as morphological transformation. Some effects, such as those on goblet cells and ciliated cells, persisted following the 7-day recovery period. Of note, the functional and structural disturbances observed in GA-treated cultures resemble those found in ortho-phthaldehyde (OPA)-treated cultures. Furthermore, our in vitro findings on GA toxicity partially and qualitatively mimicked those reported in the animal and human survey studies. Taken together, observations from this study demonstrate that the human air-liquid-interface (ALI) airway tissue model, integrated with an in vitro exposure system that simulates human inhalation exposure, could be used for in vitro-based human hazard identification and the risk characterization of aerosolized chemicals.

Ciliary Beat Frequency: Proceedings and Recommendations from a Multi-laboratory Ring Trial Using 3-D Reconstituted Human Airway Epithelium to Model Mucociliary Clearance.

The use of reconstituted human airway (RHuA) epithelial tissues to assess functional endpoints is highly relevant in respiratory toxicology, but standardised methods are lacking. In June 2015, the Institute for In Vitro Sciences (IIVS) held a technical workshop to evaluate the potential for standardisation of methods, including ciliary beat frequency (CBF). The applicability of a protocol suggested in the workshop was assessed in a multi-laboratory ring study. This report summarises the findings, and uses the similarities and differences identified between the laboratories to make recommendations for researchers in the absence of a validated method. Two software platforms for the assessment of CBF were used - Sisson-Ammons Video Analysis (SAVA; Ammons Engineering, Clio, MI, USA) and ciliaFA (National Institutes of Health, Bethesda, MD, USA). Both were utilised for multiple read temperatures, one objective strength (10×) and up to four video captures per tissue, to assess their utility. Two commercial RHuA tissue cultures were used: MucilAir™ (Epithelix, Geneva, Switzerland) and EpiAirway™ (MatTek, Ashland, MA, USA). IL-13 and procaterol were used to induce CBF-specific responses as positive controls. Further testing addressed the impact of tissue acclimation duration, the number of capture fields and objective strengths on baseline CBF readings. Both SAVA and ciliaFA reliably collected CBF data. However, ciliaFA failed to generate accurate CBF measurements above ∼10 Hz. The positive controls were effective, but were subject to inter-laboratory variability. CBF endpoints were generally uniform across replicate tissues, objective strengths and laboratories. Longer tissue acclimation increased the percentage active area, but had minimal impact on CBF. Taken together, these findings support the development and validation of a standardised CBF measurement protocol.

Porous Carbon Spheres Derived from Hemicelluloses for Supercapacitor Application.

With the increasing demand for dissolving pulp, large quantities of hemicelluloses were generated and abandoned. These hemicelluloses are very promising biomass resources for preparing carbon spheres. However, the pore structures of the carbon spheres obtained from biomass are usually poor, which extensively limits their utilization. Herein, the carbon microspheres derived from hemicelluloses were prepared using hydrothermal carbonization and further activated with different activators (KOH, K2CO3, Na2CO3, and ZnCl2) to improve their electrochemical performance as supercapacitors. After activation, the specific surface areas of these carbon spheres were improved significantly, which were in the order of ZnCl2 > K2CO3 > KOH > Na2CO3. The carbon spheres with high surface area of 2025 m2/g and remarkable pore volume of 1.07 cm3/g were achieved, as the carbon spheres were activated by ZnCl2. The supercapacitor electrode fabricated from the ZnCl2-activated carbon spheres demonstrated high specific capacitance of 218 F/g at 0.2 A/g in 6 M KOH in a three-electrode system. A symmetric supercapacitor was assembled in 2 M Li2SO4 electrolyte, and the carbon spheres activated by ZnCl2 showed excellent electrochemical performance with high specific capacitance (137 F/g at 0.5 A/g), energy densities (15.4 Wh/kg), and good cyclic stability (95% capacitance retention over 2000 cycles).

Integrated CNN and Federated Learning for COVID-19 Detection on Chest X-Ray Images.

Currently, Coronavirus Disease 2019 (COVID-19) is still endangering world health and safety and deep learning (DL) is expected to be the most powerful method for efficient detection of COVID-19. However, patients' privacy concerns prohibit data sharing between medical institutions, leading to unexpected performance of deep neural network (DNN) models. Fortunately, federated learning (FL), as a novel paradigm, allows participating clients to collaboratively train models without exposing source data outside original location. Nevertheless, the current FL-based COVID-19 detection methods prefer optimizing secondary objectives including delay, energy consumption and privacy, while few works focus on improving the model accuracy and stability. In this paper, we propose a federated learning framework with dynamic focus for COVID-19 detection on CXR images, named FedFocus. Specifically, to improve the training efficiency and accuracy, the training loss of each model is taken as the basis for parameter aggregation weights. As training layer deepens, a constantly updated dynamic factor is designed to stabilize the aggregation process. In addition, to highly restore the real dataset, the training sets in our experiments are divided based on the population and the infection of three real cities. Extensive experiments conducted on the real-world CXR images dataset demonstrate that FedFocus outperforms the baselines in model training efficiency, accuracy and stability.

Structure of corn bran hemicelluloses isolated with aqueous ethanol solutions and their potential to produce furfural.

The alkali-soluble hemicelluloses extracted with 10% KOH solution from corn bran were further isolated with different concentrations of aqueous ethanol solutions. Herein 92.2% of the original hemicelluloses can be obtained and the cellulase enzymatic hydrolysis rate of the alkali treated corn bran can reach to 97.2%. The corn bran hemicelluloses were mainly glucuronoarabinoxylan, in which xylose (48.4-53.8%) and arabinose (27.8-33.2%) were the main components. More linear hemicelluloses with high molecular weight tended to be precipitated in low concentration aqueous ethanol solutions. Furthermore, the relationship between the structural features of these alkali-soluble corn bran hemicelluloses and their furfural yield was investigated in MIBK (methyl isobutyl ketone)/H2O biphasic system. Results showed that the hemicelluloses with high xylose content are benefit to the furfural production, and the highest furfural yield of 67.7% was obtained.

Ultrafast alkaline deep eutectic solvent pretreatment for enhancing enzymatic saccharification and lignin fractionation from industrial xylose residue.

An aspirational pretreatment method for efficient fractionation and tailored valorization of large industrial biomass can ensure the realizability of sustainable biorefinery strategies. In this study, an ultrafast alkaline deep eutectic solvents (DES) pretreatment strategy was developed to efficiently extract the lignin nanoparticles and retain cellulose residues that could be readily enzymatic saccharified to obtain fermentative glucose for the bioenergy production from industrial xylose residue. Results showed that the DES pretreatment had excellent delignification performance and the regenerated DES lignin nanoparticles exhibited well-preserved structures and excellent antioxidant activity, as well as low molecular weights and relatively uniform size distribution, which could facilitate downstream catalytic degradation for production of chemicals and preparation of lignin-based materials. Under the optimal condition (DES pretreatment: 80 °C, 10 min; saccharification: 10 FPU/g, 5 wt%, 100 mg/g Tween 80), the glucose yield of 90.12% could be achieved, which was dramatically increased compared to raw materials.

Evaluating the Sub-Acute Toxicity of Formaldehyde Fumes in an In Vitro Human Airway Epithelial Tissue Model.

Formaldehyde (FA) is an irritating, highly reactive aldehyde that is widely regarded as an asthmagen. In addition to its use in industrial applications and being a product of combustion reaction and endogenous metabolism, FDA-regulated products may contain FA or release FA fumes that present toxicity risks for both patients and healthcare workers. Exposure to airborne FA is associated with nasal neoplastic lesions in both animals and humans. It is classified as a Group 1 carcinogen by International Agency for Research on Cancer (IARC) based on the increased incidence of cancer in animals and a known human carcinogen in the Report on Carcinogens by National Toxicology Program (NTP). Herein, we systematically evaluated the tissue responses to FA fumes in an in vitro human air-liquid-interface (ALI) airway tissue model. Cultures were exposed at the air interface to 7.5, 15, and 30 ppm of FA fumes 4 h per day for 5 consecutive days. Exposure to 30 ppm of FA induced sustained oxidative stress, along with functional changes in ciliated and goblet cells as well as possible squamous differentiation. Furthermore, secretion of the proinflammatory cytokines, IL-1ß, IL-2, IL-8, GM-CSF, TNF-a and IFN-γ, was induced by repeated exposures to FA fumes. Expression of MMP-1, MMP-3, MMP-7, MMP-10, MMP-12, and MMP-13 was downregulated at the end of the 5-day exposure. Although DNA-damage was not detected by the comet assay, FA exposures downregulated the DNA repair enzymes MGMT and FANCD2, suggesting its possible interference in the DNA repair capacity. Overall, a general concordance was observed between our in vitro responses to FA fume exposures and the reported in vivo toxicity of FA. Our findings provide further evidence supporting the application of the ALI airway system as a potential in vitro alternative for screening and evaluating the respiratory toxicity of inhaled substances.

Exploration of deep eutectic solvent-based biphasic system for furfural production and enhancing enzymatic hydrolysis: Chemical, topochemical, and morphological changes.

Developing a biorefinery process for a highly integrated valorization and fractionation of lignocellulose is crucial for its utilization. Herein, a biphasic system comprising choline chloride/lactic acid and 2-methyltetrahydrofuran with Al2(SO4)3 and H2SO4 as catalysts was applied to pretreat Eucalyptus. Results showed that under the optimized conditions (150 °C, 30 min, 0.2 M Al2(SO4)3, 0.075 M H2SO4), the furfural yield and enzymatic hydrolysis efficiency could reach 54.7% and 97.0%, respectively. The efficient cellulose conversion was attributed to remarkable removal of lignin (91.0%) and hemicelluloses (100.0%), thereby causing the disruption of cell wall structure and enhancement of cellulose accessibility. Meanwhile, confocal Raman microscope and atomic force microscope displayed that the pretreatment resulted in the decreasing intensities of carbohydrates and lignin different regions of cell walls, and exposing of the embedded microfibers from noncellulosic polymers. Overall, the deep eutectic solvent-based biphasic system displayed high performance for effective utilization of carbohydrate components in lignocellulose.

Hsa_circ_0007059 promotes apoptosis and inflammation in cardiomyocytes during ischemia by targeting microRNA-378 and microRNA-383.

Circular RNAs (circRNAs) are a class of non-coding RNA molecules that are associated with not only normal physiological functions but also various diseases, including cardiac diseases such as myocardial infarction (MI). The present study explored the potential role of circRNA_0007059 (circ_0007059) during MI pathogenesis using in vitro studies. Microarray and quantitative PCR analyses demonstrated elevated circ_0007059 expression and downregulated miR-378 and miR-383 expression in H2O2-treated mice cardiomyocytes and infarcted hearts of MI mouse model as compared those in relevant controls. Moreover, circ_0007059 knockdown improved cardiomyocyte viability after H2O2 treatment as revealed by the CCK-8 and colony formation assays. Flow cytometry and caspase activity assays demonstrated that circ_0007059 suppressed H2O2-induced cardiomyocyte apoptosis. Enzyme-linked immunosorbent assays and Western blotting revealed that inflammatory cytokine (interleukin-1ß, interleukin-18 and C-C motif chemokine ligand 5) expression was induced by H2O2 treatment and that circ_0007059 repressed H2O2-induced inflammation. Bioinformatics analyses and dual-luciferase reporter assays showed that circ_0000759 acts as a miR-378 and miR-383 sponge. Furthermore, the upregulation or suppression of miR-378 and miR-383 expression in H2O2-treated cardiomyocytes had similar effects on the apoptosis and inflammation of cardiomyocytes as that of circ_0007059 knockdown or overexpression, respectively. Additionally, lentiviral shRNA-circ_0007059 administration to mice with MI considerably reduced the size of infarcted regions and promoted cardiac activity. Collectively, our findings suggest that circ_0007059 expression is upregulated in mice cardiomyocytes in response to oxidative stress and cardiac tissues of MI mouse model, suggesting its involvement in the pathogenesis of MI by targeting miR-378 and miR-383.

Infinite Bayesian Max-Margin Discriminant Projection.

In this article, considering the supervised dimensionality reduction, we first propose a model, called infinite Bayesian max-margin linear discriminant projection (iMMLDP), by assembling a set of local regions, where we make use of Bayesian nonparametric priors to handle the model selection problem, for example, the underlying number of local regions. In each local region, our model jointly learns a discriminative subspace and the corresponding classifier. Under this framework, iMMLDP combines dimensionality reduction, clustering, and classification in a principled way. Moreover, to deal with more complex data, for example, a local nonlinear separable structure, we extend the linear projection to a nonlinear case based on the kernel trick and develop an infinite kernel max-margin discriminant projection (iKMMDP) model. Thanks to the conjugate property, the parameters in these two models can be inferred efficiently via the Gibbs sampler. Finally, we implement our models on synthesized and real-world data, including multimodally distributed datasets and measured radar image data, to validate their efficiency and effectiveness.

Max-Margin Deep Diverse Latent Dirichlet Allocation With Continual Learning.

Deep probabilistic aspect models are widely utilized in document analysis to extract the semantic information and obtain descriptive topics. However, there are two problems that may affect their applications. One is that common words shared among all documents with low representational meaning may reduce the representation ability of learned topics. The other is introducing supervision information to hierarchical topic models to fully utilize the side information of documents that is difficult. To address these problems, in this article, we first propose deep diverse latent Dirichlet allocation (DDLDA), a deep hierarchical topic model that can yield more meaningful semantic topics with less common and meaningless words by introducing shared topics. Moreover, we develop a variational inference network for DDLDA, which helps us to further generalize DDLDA to a supervised deep topic model called max-margin DDLDA (mmDDLDA) by employing max-margin principle as the classification criterion. Compared to DDLDA, mmDDLDA can discover more discriminative topical representations. In addition, a continual hybrid method with stochastic-gradient MCMC and variational inference is put forward for deep latent Dirichlet allocation (DLDA)-based models to make them more practical in real-world applications. The experimental results demonstrate that DDLDA and mmDDLDA are more efficient than existing unsupervised and supervised topic models in discovering highly discriminative topic representations and achieving higher classification accuracy. Meanwhile, DLDA and our proposed models trained by the proposed continual learning approach cannot only show good performance on preventing catastrophic forgetting but also fit the evolving new tasks well.

Effect of life stage and target tissue on dose-response assessment of ethyl methane sulfonate-induced genotoxicity.

In order to investigate the possibility that treatment age affects the genotoxic response to ethyl methane sulfonate (EMS) exposure, we dosed gpt-delta neonatal mice on postnatal days 1-28 with 5-100 mg/kg/day of EMS and measured micronucleus (MN) induction in peripheral blood and gpt gene mutation in liver, lung, bone marrow, small intestine, spleen, and kidney. The data were compared to measurements from similarly exposed adult gpt-delta mice. Our results indicate that the peripheral blood MN frequencies in mice treated as neonates are not substantially different from those measured in mice treated as adults. There were, however, differences in tissue-specific gpt mutation responses in mice treated with EMS as neonates and adults. Greater mutant frequencies were seen in DNA isolated from kidney of mice treated as neonates, whereas the mutant frequencies in bone marrow, liver, and spleen were greater in the animals treated as adults. Benchmark dose potency ranking indicated that the differences for kidney were significant. Our data indicate that there are differences in EMS-induced genotoxicity between mice treated as adults and neonates; the differences, however, are relatively small.