Notes

An incident record of bilateral pseudo-doubling with the optic disks.
Recent methods in network pruning have indicated that a dense neural network involves a sparse subnetwork (called a winning ticket), which can achieve similar test accuracy to its dense counterpart with much fewer network parameters. Generally, these methods search for the winning tickets on well-labeled data. Unfortunately, in many real-world applications, the training data are unavoidably contaminated with noisy labels, thereby leading to performance deterioration of these methods. To address the above-mentioned problem, we propose a novel two-stream sample selection network (TS 3 -Net), which consists of a sparse subnetwork and a dense subnetwork, to effectively identify the winning ticket with noisy labels. The training of TS 3 -Net contains an iterative procedure that switches between training both subnetworks and pruning the smallest magnitude weights of the sparse subnetwork. In particular, we develop a multistage learning framework including a warm-up stage, a semisupervised alternate learning stage, and a label refinement stage, to progressively train the two subnetworks. In this way, the classification capability of the sparse subnetwork can be gradually improved at a high sparsity level. Extensive experimental results on both synthetic and real-world noisy datasets (including MNIST, CIFAR-10, CIFAR-100, ANIMAL-10N, Clothing1M, and WebVision) demonstrate that our proposed method achieves state-of-the-art performance with very small memory consumption for label noise learning. Code is available at https//github.com/Runqing-forMost/TS3-Net/tree/master.Reaching and maintaining high walking speeds is challenging for a human when carrying extra weight, such as walking with a heavy backpack. Robotic limbs can support a heavy backpack when standing still, but accelerating a backpack within a couple of steps to race-walking speeds requires limb force and energy beyond natural human ability. Here, we conceive a human-driven robot exoskeleton that could accelerate a heavy backpack faster and maintain top speeds higher than what the human alone can when not carrying a backpack. check details The key components of the exoskeleton are the mechanically adaptive but energetically passive spring limbs. We show that by optimally adapting the stiffness of the limbs, the robot can achieve near-horizontal center of mass motion to emulate the load-bearing mechanics of the bicycle. We find that such an exoskeleton could enable the human to accelerate one extra body weight up to top race-walking speeds in ten steps. Our finding predicts that human-driven mechanically adaptive robot exoskeletons could extend human weight-bearing and fast-walking ability without using external energy.Electromyography (EMG) is one of the most common methods to detect muscle activities and intentions. However, it has been difficult to estimate accurate hand motions represented by the finger joint angles using EMG signals. We propose an encoder-decoder network with an attention mechanism, an explainable deep learning model that estimates 14 finger joint angles from forearm EMG signals. This study demonstrates that the model trained by the single-finger motion data can be generalized to estimate complex motions of random fingers. The color map result of the after-training attention matrix shows that the proposed attention algorithm enables the model to learn the nonlinear relationship between the EMG signals and the finger joint angles, which is explainable. The highly activated entries in the color map of the attention matrix derived from model training are consistent with the experimental observations in which certain EMG sensors are highly activated when a particular finger moves. In summary, this study proposes an explainable deep learning model that estimates finger joint angles based on EMG signals of the forearm using the attention mechanism.Biologically important effects occur when proteins bind to other substances, of which binding to DNA is a crucial one. Therefore, accurate identification of protein-DNA binding residues is important for further understanding of the protein-DNA interaction mechanism. Although wet-lab methods can accurately obtain the location of bound residues, it requires significant human, financial and time costs. There is thus an urgent need to develop efficient computational-based methods. Most current state-of-the-art methods are two-step approaches the first step uses a sliding window technique to extract residue features; the second step uses each residue as an input to the model for prediction. This has a negative impact on the efficiency of prediction and ease of use. In this study, we propose a sequence-to-sequence (seq2seq) model that can input the entire protein sequence of variable length and use two modules, Transformer Encoder Block and Feature Extracting Block, for hierarchical feature extraction, where Transformer Encoder Block is used to extract global features, and then Feature Extracting Block is used to extract local features to further improve the recognition capability of the model. The comparison results on two benchmark datasets, namely PDNA-543 and PDNA-41, prove the effectiveness of our method in identifying protein-DNA binding residues. The code is available at https//github.com/ShixuanGG/DNA-protein_binding_residues.Daphnia, an ecologically important zooplankton species in lakes, shows both genetic adaptation and phenotypic plasticity in response to temperature and fish predation, but little is known about the molecular basis of these responses and their potential interactions. We performed a factorial experiment exposing laboratory-propagated Daphnia pulicaria clones from two lakes in the Sierra Nevada mountains of California to normal or high temperature (15°C or 25°C) in the presence or absence of fish kairomones, then measured changes in life history and gene expression. Exposure to kairomones increased upper thermal tolerance limits for physiological activity in both clones. Cloned individuals matured at a younger age in response to higher temperature and kairomones, while size at maturity, fecundity and population intrinsic growth were only affected by temperature. At the molecular level, both clones expressed more genes differently in response to temperature than predation, but specific genes involved in metabolic, cellular, and genetic processes responded differently between the two clones. Although gene expression differed more between clones from different lakes than experimental treatments, similar phenotypic responses to predation risk and warming arose from these clone-specific patterns. Our results suggest that phenotypic plasticity responses to temperature and kairomones interact synergistically, with exposure to fish predators increasing the tolerance of Daphnia pulicaria to stressful temperatures, and that similar phenotypic responses to temperature and predator cues can be produced by divergent patterns of gene regulation.The HTLV-1 protease is one of the major antiviral targets to overwhelm this virus. Several research groups have developed protease inhibitors, but none has been successful. In this regard, developing new HTLV-1 protease inhibitors to fix the defects in previous inhibitors may overcome the lack of curative treatment for this oncovirus. Thus, we decided to study the unbinding pathways of the most potent (compound 10, PDB ID 4YDF, Ki = 15 nM) and one of the weakest (compound 9, PDB ID 4YDG, Ki = 7900 nM) protease inhibitors, which are very structurally similar. We conducted 12 successful short and long simulations (totaling 14.8 μs) to unbind the compounds from two monoprotonated (mp) forms of protease using the Supervised Molecular Dynamics (SuMD) without applying any biasing force. The results revealed that Asp32 or Asp32' in the two forms of mp state similarly exert powerful effects on maintaining both potent and weak inhibitors in the binding pocket of HTLV-1 protease. In the potent inhibitor's unbinding process, His66' was a great supporter that was absent in the weak inhibitor's unbinding pathway. In contrast, in the weak inhibitor's unbinding process, Trp98/Trp98' by pi-pi stacking interactions were unfavorable for the stability of the inhibitor in the binding site. In our opinion, these results will assist in designing more potent and effective inhibitors for the HTLV-1 protease.The bovine virus diarrhea virus (BVDV) causes reproductive, enteric, and respiratory diseases. Vaccination is essential in increasing herd resistance to BVDV spread. The selection of an adjuvant is an important factor in the success of the vaccination process. Monolaurin or glycerol monolaurate is a safe compound with an immunomodulatory effect. This study aimed to evaluate the efficacy of monolaurin as a novel adjuvant. This was examined through the preparation of an inactivated BVDV (NADL strain) vaccine adjuvanted with different concentrations of monolaurin and compared with the registered available locally prepared polyvalent vaccine (Pneumo-4) containing BVD (NADL strain), BoHV-1 (Abou Hammad strain), BPI3 (strain 45), and BRSV (strain 375L), and adjuvanted with aluminum hydroxide gel. The inactivated BVDV vaccine was prepared using three concentrations, 0.5%, 1%, and 2%, from monolaurin as adjuvants. A potency test was performed on five groups of animals. The first group, which did not receive vaccination, served as a control group while three other groups were vaccinated using the prepared vaccines. The fifth group received the Pneumo-4 vaccine. Vaccination response was monitored by measuring viral neutralizing antibodies using enzyme-linked immunosorbent assay (ELISA). It was found that the BVD inactivated vaccine with 1% and 2% monolaurin elicited higher neutralizing antibodies that have longer-lasting effects (nine months) with no reaction at the injection site in comparison to the commercial vaccine adjuvanted by aluminum hydroxide gel.Genetic influences on body mass index (BMI) appear to markedly differ across life, yet existing research is equivocal and limited by a paucity of life course data. We thus used a birth cohort study to investigate differences in association and explained variance in polygenic risk for high BMI across infancy to old age (2-69 years). A secondary aim was to investigate how the association between BMI and a key purported environmental determinant (childhood socioeconomic position) differed across life, and whether this operated independently and/or multiplicatively of genetic influences. Data were from up to 2677 participants in the MRC National Survey of Health and Development, with measured BMI at 12 timepoints from 2-69 years. We used multiple polygenic indices from GWAS of adult and childhood BMI, and investigated their associations with BMI at each age. For polygenic liability to higher adult BMI, the trajectories of effect size (β) and explained variance (R2) diverged explained variance peaked in early adulthood and plateaued thereafter, while absolute effect sizes increased throughout adulthood. For polygenic liability to higher childhood BMI, explained variance was largest in adolescence and early adulthood; effect sizes were marginally smaller in absolute terms from adolescence to adulthood. All polygenic indices were related to higher variation in BMI; quantile regression analyses showed that effect sizes were sizably larger at the upper end of the BMI distribution. Socioeconomic and polygenic risk for higher BMI across life appear to operate additively; we found little evidence of interaction. Our findings highlight the likely independent influences of polygenic and socioeconomic factors on BMI across life. Despite sizable associations, the BMI variance explained by each plateaued or declined across adulthood while BMI variance itself increased. This is suggestive of the increasing importance of chance ('non-shared') environmental influences on BMI across life.
My Website: https://www.selleckchem.com/products/trastuzumab-emtansine-t-dm1-.html