Estimating Energy Concentrations in Wooded Pastures of NW Spain Using Empirical Models That Relate Observed Metabolizable Energy to Measured Nutritional Attributes

Wooded pastures serve as a traditional source of forage in Europe, where forest grazing is valued as an efficient tool for maintaining the diversity of semi-natural habitats. In a forest grazing setting with diverse diet composition, assessing the energy content of animal diets can be a difficult task because of its dependency on digestibility measures. In the present study, prediction equations of metabolizable energy (ME) were obtained performing stepwise regression with data (n = 297; 44 plant species) on nutritional attributes (Acid Detergent Fiber, lignin, silica, dry matter, crude protein, in vitro organic matter digestibility) from 20 representative stands of Atlantic dry heathlands and pedunculate oak woodlands. The results showed that the prediction accuracy of ME is reduced when the general model (R2 = 0.64) is applied, as opposed to the use of the specific prediction equations for each vegetation type (R2 = 0.61, 0.66, 0.71 for oak woodlands; R2 = 0.70 heather-gorse dominated heathlands, R2 = 0.41 continental heathlands). The general model tends to overestimate the ME concentrations in heaths with respect to the observed ME values obtained from IVOMD as a sole predictor, and this divergence could be corrected by applying the specific prediction equations obtained for each vegetation type. Although the use of prediction equations by season would improve accuracy in the case of a Winter scenario, using the general model as opposed to the prediction equations for Spring, Summer or Fall would represent a much smaller loss of accuracy.

Investigation of accuracy of rain-rate and rain-attenuation prediction models in satellite communications based on meteorological skills

Rainfall is a major destructive factor which severely reduces the quality and reliability of propagated signals in satellite communications. Hence, rain-attenuation prediction plays a vital role in the satellite radio link planning and engineering. The accuracy of the rain-attenuation prediction models depends on two things; (i) the accuracy of rain-rate information and (ii) the area of study. Therefore, selecting an appropriate rain-attenuation prediction model for a new site without having any specific prediction model and experimental measured rain-rate would be challenging. In this regard, this letter takes advantage of climatology skills to find an accurate model for such kind of areas. To do so, we study the Urmia-site (37.55° N, 45.1° E) and its communication link with the Eutelsat 25A (25.5° E), where there is no available experimental measured data and specific prediction models for that site. Therefore, based on the meteorological skills, the Yong-in site in South-Korea (37.43° N, 126.93° E) was chosen, as a homogeneous area with Urmia, which has available measured data of rainfall and rain-attenuation. Afterward, the most common used global prediction models are applied to Yong-in and the results are compared with the existing measurements. Consequently, the more accurate rain-rate and rain-attenuation prediction models are investigated and generalized to Urmia, which are the ITU-R P.837-5 model with 34% r.m.s. and the Joo-Hwan model with 18% r.m.s., respectively. Finally, the amount of rain-attenuation in different useful frequency bands (10-50 GHz) is investigated for Urmia by the Joo-Hwan model.

Utilisation of ‘Aerostat’ Doppler Weather Radar in nowcasting of convective phenomena

‘Aerostat’ system is a part of the air defence radar network, adopted by the Indian Air Force. Many meteorological instruments have been integrated with this system, including Doppler Weather Radar (DWR). The ground-based DWR has a maximum range of 300 NM, however, it generally uses 150 NM range on scan mode. The scan mode images are provided at half an hour interval, which are being utilised very effectively for nowcasting of thunderstorms at various IAF bases. In the present study, utilisation of DWR images for nowcasting of thunderstorms / dust storms is discussed over NW India with the help of a few case studies during pre-monsoon and SW monsoon seasons of 2008. Further, products generated through operational meso-scale NWP model runs have been studied in order to obtain indications / guidance for expected convective activity over the area at least 24-36 hours in advance. Thus, short-range weather forecasts through NWP models can be used as an advance indication for careful monitoring of DWR images in near real time. It has been found that the DWR is a very good tool to track the movement of significant weather echoes around the airfields, which can be very helpful in issuing appropriate warnings / advisories with sufficient lead time. Meso-scale NWP models are capable of generating reliable indications for expected convective activity at least 24-36 hours in advance. The integration of both the inputs can increase the accuracy and reliability of location and time specific prediction of convective activity.

Comparative epigenetic analysis of tumour initiating cells and syngeneic EPSC-derived neural stem cells in glioblastoma

Epigenetic mechanisms which play an essential role in normal developmental processes, such as self-renewal and fate specification of neural stem cells (NSC) are also responsible for some of the changes in the glioblastoma (GBM) genome. Here we develop a strategy to compare the epigenetic and transcriptional make-up of primary GBM cells (GIC) with patient-matched expanded potential stem cell (EPSC)-derived NSC (iNSC). Using a comparative analysis of the transcriptome of syngeneic GIC/iNSC pairs, we identify a glycosaminoglycan (GAG)-mediated mechanism of recruitment of regulatory T cells (Tregs) in GBM. Integrated analysis of the transcriptome and DNA methylome of GBM cells identifies druggable target genes and patient-specific prediction of drug response in primary GIC cultures, which is validated in 3D and in vivo models. Taken together, we provide a proof of principle that this experimental pipeline has the potential to identify patient-specific disease mechanisms and druggable targets in GBM.

Multi-tissue proteomics identifies molecular signatures for sporadic and genetically defined Alzheimer disease cases

Alzheimer disease (AD) is a heterogeneous disease with many genes are associated with AD risk. Most proteomic studies, while instrumental in identifying AD pathways and genes, focus on single tissues and sporadic AD cases. Multi-tissue proteomic signatures for sporadic and genetically defined AD (e.g., pathogenic variant carriers in APP and PSEN1/2 and risk variant carriers in TREM2) will illuminate the biology of this heterogeneous disease.1,2 Here, we present one of the largest multi-tissue proteomic profiles, accessible through our web portal, based on 1,305 proteins in brain (n=360), cerebrospinal fluid (CSF; n=717), and plasma (n=490) from the Knight Alzheimer Disease Research Center (Knight ADRC) and Dominantly Inherited Alzheimer Network (DIAN) cohorts.3-5 We identified proteomic signatures in brain, CSF, and plasma for sporadic AD status and replicated these findings in multiple, independent datasets. The area under the curve (AUC) for CSF proteins was 0.89 in discovery and 0.90 in the replication dataset, which was significantly higher than the AUC for CSF p-tau181/Aβ42 (AUC = 0.81; P = 2.4×10-6). We also identified a specific proteomic signature for TREM2 variant carriers that differentiated TREM2 variant carriers from sporadic AD cases and controls with high sensitivity and specificity (AUC = 0.81 - 1). In addition, the proteins that showed differential levels in sporadic AD were also altered in autosomal dominant AD, but with greater effect size (1.4 times, P = 3.8×10-5), and proteins associated with autosomal dominant AD, in brain tissue also replicated on CSF (p=1.36×10-9). Enrichment analyses highlighted several pathways including AD (calcineurin, APOE, GRN), Parkinson disease (α-synuclein, LRRK2), and innate immune response (SHC1, MAPK3, SPP1) for the sporadic AD or TREM2 variant carriers. Our findings show the power of multi-tissue proteomics’ contribution to the understanding of AD biology and to the creation of tissue-specific prediction models for individuals with specific genetic profiles, ultimately supporting its utility in creating individualized disease risk evaluation and treatment.

Universal Adversarial Attack via Conditional Sampling for Text Classification

Despite deep neural networks (DNNs) having achieved impressive performance in various domains, it has been revealed that DNNs are vulnerable in the face of adversarial examples, which are maliciously crafted by adding human-imperceptible perturbations to an original sample to cause the wrong output by the DNNs. Encouraged by numerous researches on adversarial examples for computer vision, there has been growing interest in designing adversarial attacks for Natural Language Processing (NLP) tasks. However, the adversarial attacking for NLP is challenging because text is discrete data and a small perturbation can bring a notable shift to the original input. In this paper, we propose a novel method, based on conditional BERT sampling with multiple standards, for generating universal adversarial perturbations: input-agnostic of words that can be concatenated to any input in order to produce a specific prediction. Our universal adversarial attack can create an appearance closer to natural phrases and yet fool sentiment classifiers when added to benign inputs. Based on automatic detection metrics and human evaluations, the adversarial attack we developed dramatically reduces the accuracy of the model on classification tasks, and the trigger is less easily distinguished from natural text. Experimental results demonstrate that our method crafts more high-quality adversarial examples as compared to baseline methods. Further experiments show that our method has high transferability. Our goal is to prove that adversarial attacks are more difficult to detect than previously thought and enable appropriate defenses.

Development of the Hypertension Index Model in General Adult Using the Korea National Health and Nutritional Examination Survey and the Korean Genome and Epidemiology Study

Hypertension, a risk factor for cardiovascular disease and all-cause mortality, has been increasing. Along with emphasizing awareness and control of hypertension, predicting the incidence of hypertension is important. Several studies have previously reported prediction models of hypertension. However, among the previous models for predicting hypertension, few models reflect various risk factors for hypertension. We constructed a sex-specific prediction model using Korean datasets, which included socioeconomic status, medical history, lifestyle-related variables, anthropometric status, and laboratory indices. We utilized the data from the Korea National Health and Nutrition Examination Survey from 2011 to 2015 to derive a hypertension prediction model. Participants aged 40 years or older. We constructed a sex-specific hypertension classification model using logistic regression and features obtained by literature review and statistical analysis. We constructed a sex-specific hypertension classification model including approximately 20 variables. We estimated its performance using the Korea National Health and Nutrition Examination Survey dataset from 2016 to 2018 (AUC = 0.847 in men, AUC = 0.901 in women). The performance of our hypertension model was considered significant based on the cumulative incidence calculated from a longitudinal dataset, the Korean Genome and Epidemiology Study dataset. We developed this hypertension prediction model using features that could be collected in a clinical office without difficulty. Individualized results may alert a person at high risk to modify unhealthy lifestyles.