This is a PDF file of a peer-reviewed paper that has been accepted for publication. Although unedited, the content has been subjected to preliminary formatting. Nature is providing this early version of the typeset paper as a service to our authors and readers. The text and figures will undergo copyediting and a proof review before the paper is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers apply.

The search for earth-abundant and high-performance electrode materials for sodium-ion batteries This article is protected by copyright. All rights reserved.2 represents an important challenge to current battery research. Two-dimensional transition metal dichalcogenides, particularly MoS 2 , have attracted increasing attention recently, but few of them so far have been able to meet expectations. In this study, we demonstrate that another phase of molybdenum sulfide -amorphous chain-like MoS 3 -can be a better choice as the anode material of sodium-ion batteries. We prepare highly compact MoS 3 particles infiltrated with carbon nanotubes via the facile acid precipitation method in ethylene glycol. Compared to crystalline MoS 2 , the resultant amorphous MoS 3 not only exhibits impressive gravimetric performance -featuring excellent specific capacity (~615 mAh/g), rate capability (235 mAh/g at 20 A/g) and cycling stability, but also shows exceptional volumetric capacity of ~1000 mAh/cm 3 and an areal capacity of >6.0 mAh/cm 2 at very high areal loadings of active materials (up to 12 mg/cm 2 ). Our experimental results are supported by DFT simulations showing that the 1D chains of MoS 3 can facilitate the adsorption and diffusion of Na + ions. At last, we demonstrate that the MoS 3 anode can be paired with a Na 3 V 2 (PO 4 ) 3 cathode to afford full cells with great capacity and cycling performance.

Neuropsychological deficits, including impairments in learning and memory, occur after spinal cord injury (SCI). In experimental SCI models, we and others have reported that such changes reflect sustained microglia activation in the brain that is associated with progressive neurodegeneration. In the present study, we examined the effect of pharmacological depletion of microglia on posttraumatic cognition, depressive-like behavior, and brain pathology after SCI in mice. Methods: Young adult male C57BL/6 mice were subjected to moderate/severe thoracic spinal cord contusion. Microglial depletion was induced with the colony-stimulating factor 1 receptor (CSF1R) antagonist PLX5622 administered starting either 3 weeks before injury or one day post-injury and continuing through 6 weeks after SCI. Neuroinflammation in the injured spinal cord and brain was assessed using flow cytometry and NanoString technology. Neurological function was evaluated using a battery of neurobehavioral tests including motor function, cognition, and depression. Lesion volume and neuronal counts were quantified by unbiased stereology. Results: Flow cytometry analysis demonstrated that PLX5622 pre-treatment significantly reduced the number of microglia, as well as infiltrating monocytes and neutrophils, and decreased reactive oxygen species production in these cells from injured spinal cord at 2-days post-injury. Post-injury PLX5622 treatment reduced both CD45 int microglia and CD45 hi myeloid counts at 7-days. Following six weeks of PLX5622 treatment, there were substantial changes in the spinal cord and brain transcriptomes, including those involved in neuroinflammation. These alterations were associated with improved neuronal survival in the brain and neurological recovery. Conclusion: These findings indicate that pharmacological microglia-deletion reduces neuroinflammation in the injured spinal cord and brain, improving recovery of cognition, depressive-like behavior, and motor function.

The scientific community is rapidly generating protein sequence information, but only a fraction of these proteins can be experimentally characterized. While promising deep learning approaches for protein prediction tasks have emerged, they have computational limitations or are designed to solve a specific task. We present a Transformer neural network that pre-trains task-agnostic sequence representations. This model is fine-tuned to solve two different protein prediction tasks: protein family classification and protein interaction prediction. Our method is comparable to existing state-of-the art approaches for protein family classification, while being much more general than other architectures. Further, our method outperforms all other approaches for protein interaction prediction. These results offer a promising framework for fine-tuning the pre-trained sequence representations for other protein prediction tasks.to identify functional characteristics is critical to understanding cellular functions as well as developing potential therapeutic applications [4]. Sequence-based methods to computationally infer protein characteristics have been critical for inferring protein function and other characteristics [5]. Thus, the development of computational methods to infer protein characteristics (which we generally describe as "protein prediction tasks") has become paramount in the field of bioinformatics and computational biology. Here, we develop a Transformer neural network to establish task-agnostic representations of protein sequences, and use the Transformer network to solve two protein prediction tasks. Background: Deep LearningDeep learning, a class of machine learning based on the use of artificial neural networks, has recently transformed the field of computational biology and medicine through its application towards long-standing problems such as image analysis, gene expression modeling, sequence variant calling, and putative drug discovery [6,7,8,9,10]. By leveraging deep learning, field specialists have been able to efficiently design and train models without the extensive feature engineering required by previous methods. In applying deep learning to sequence-based protein characterization tasks, we first consider the field of natural language processing (NLP), which aims to analyze human language through computational techniques [11]. Deep learning has recently proven to be a critical tool for NLP, achieving state-ofthe-art performance on benchmarks for named entity recognition, sentiment analysis, question answering, and text summarization, among others [12,13].Neural networks are functions that map one vector space to another. Thus, in order to use them for NLP tasks, we first need to represent words as real-valued vectors. Often referred to as word embeddings, these vector representations are typically "pre-trained" on an auxiliary task for which we have (or can automatically generate) a large amount of training data. The goal of this pre-training is to learn generically useful representations that enc...

No abstract