Enhancing Domain Diversity of Transfer Learning-Based SSVEP-BCIs by the Reconstruction of Channel Correlation.

OBJECTIVE: The application of transfer learning, specifically pre-training and fine-tuning, in steady-state visual evoked potential (SSVEP)-based brain-computer interfaces (BCIs) has been demonstrated to effectively improve the classification performance of deep learning methods with limited calibration data. However, effectively learning task-related knowledge from source domains during the pre-training phase remains challenging. To address this issue, this study proposes an effective data augmentation method called Reconstruction of Channel Correlation (RCC) to optimize the utilization of the source domain data. METHODS: Concretely, RCC reconstructs training samples using probabilistically mixed eigenvector matrices derived from covariance matrices across source domains. This process manipulates the channel correlation of training samples, implicitly creating novel synthesized domains. By increasing the diversity of source domains, RCC aims to enhance the domain generalizability of the pre-trained model. The effectiveness of RCC is validated through subject-independent and subject-adaptive classification experiments. RESULTS: The results of subject-independent classification demonstrate that RCC significantly improves the classification performance of the pre-trained model on unseen target subjects. Moreover, when compared to the fine-tuning process using the RCC-absent pre-trained model, the fine-tuning process using the RCC-enhanced pre-trained model yields significantly improved performance in the subject-adaptive classification. CONCLUSION: RCC proves to enhance the performance of transfer learning by optimizing the utilization of the source domain data. SIGNIFICANCE: The RCC-enhanced transfer learning has the potential to facilitate the practical implementation of SSVEP-BCIs in real-world scenarios.

A robust ensemble deep learning framework for accurate diagnoses of tuberculosis from chest radiographs.

Introduction: Tuberculosis (TB) stands as a paramount global health concern, contributing significantly to worldwide mortality rates. Effective containment of TB requires deployment of cost-efficient screening method with limited resources. To enhance the precision of resource allocation in the global fight against TB, this research proposed chest X-ray radiography (CXR) based machine learning screening algorithms with optimization, benchmarking and tuning for the best TB subclassification tasks for clinical application. Methods: This investigation delves into the development and evaluation of a robust ensemble deep learning framework, comprising 43 distinct models, tailored for the identification of active TB cases and the categorization of their clinical subtypes. The proposed framework is essentially an ensemble model with multiple feature extractors and one of three fusion strategies-voting, attention-based, or concatenation methods-in the fusion stage before a final classification. The comprised de-identified dataset contains records of 915 active TB patients alongside 1,276 healthy controls with subtype-specific information. Thus, the realizations of our framework are capable for diagnosis with subclass identification. The subclass tags include: secondary tuberculosis/tuberculous pleurisy; non-cavity/cavity; secondary tuberculosis only/secondary tuberculosis and tuberculous pleurisy; tuberculous pleurisy only/secondary tuberculosis and tuberculous pleurisy. Results: Based on the dataset and model selection and tuning, ensemble models show their capability with self-correction capability of subclass identification with rendering robust clinical predictions. The best double-CNN-extractor model with concatenation/attention fusion strategies may potentially be the successful model for subclass tasks in real application. With visualization techniques, in-depth analysis of the ensemble model's performance across different fusion strategies are verified. Discussion: The findings underscore the potential of such ensemble approaches in augmenting TB diagnostics with subclassification. Even with limited dataset, the self-correction within the ensemble models still guarantees the accuracies to some level for potential clinical decision-making processes in TB management. Ultimately, this study shows a direction for better TB screening in the future TB response strategy.

Rare codon recoding for efficient noncanonical amino acid incorporation in mammalian cells.

The ability to genetically encode noncanonical amino acids (ncAAs) has empowered proteins with improved or previously unknown properties. However, existing strategies in mammalian cells rely on the introduction of a blank codon to incorporate ncAAs, which is inefficient and limits their widespread applications. In this study, we developed a rare codon recoding strategy that takes advantage of the relative rarity of the TCG codon to achieve highly selective and efficient ncAA incorporation through systematic engineering and big data-model predictions. We highlight the broad utility of this strategy for the incorporation of dozens of ncAAs into various functional proteins at the wild-type protein expression levels, as well as the synthesis of proteins with up to six-site ncAAs or four distinct ncAAs in mammalian cells for downstream applications.

Characterize direct protein interactions with enrichable, cleavable and latent bioreactive unnatural amino acids.

Latent bioreactive unnatural amino acids (Uaas) have been widely used in the development of covalent drugs and identification of protein interactors, such as proteins, DNA, RNA and carbohydrates. However, it is challenging to perform high-throughput identification of Uaa cross-linking products due to the complexities of protein samples and the data analysis processes. Enrichable Uaas can effectively reduce the complexities of protein samples and simplify data analysis, but few cross-linked peptides were identified from mammalian cell samples with these Uaas. Here we develop an enrichable and multiple amino acids reactive Uaa, eFSY, and demonstrate that eFSY is MS cleavable when eFSY-Lys and eFSY-His are the cross-linking products. An identification software, AixUaa is developed to decipher eFSY mass cleavable data. We systematically identify direct interactomes of Thioredoxin 1 (Trx1) and Selenoprotein M (SELM) with eFSY and AixUaa.

A Novel Data Augmentation Approach Using Mask Encoding for Deep Learning-Based Asynchronous SSVEP-BCI.

Deep learning (DL)-based methods have been successfully employed as asynchronous classification algorithms in the steady-state visual evoked potential (SSVEP)-based brain-computer interface (BCI) system. However, these methods often suffer from the limited amount of electroencephalography (EEG) data, leading to overfitting. This study proposes an effective data augmentation approach called EEG mask encoding (EEG-ME) to mitigate overfitting. EEG-ME forces models to learn more robust features by masking partial EEG data, leading to enhanced generalization capabilities of models. Three different network architectures, including an architecture integrating convolutional neural networks (CNN) with Transformer (CNN-Former), time domain-based CNN (tCNN), and a lightweight architecture (EEGNet) are utilized to validate the effectiveness of EEG-ME on publicly available benchmark and BETA datasets. The results demonstrate that EEG-ME significantly enhances the average classification accuracy of various DL-based methods with different data lengths of time windows on two public datasets. Specifically, CNN-Former, tCNN, and EEGNet achieve respective improvements of 3.18%, 1.42%, and 3.06% on the benchmark dataset as well as 11.09%, 3.12%, and 2.81% on the BETA dataset, with the 1-second time window as an example. The enhanced performance of SSVEP classification with EEG-ME promotes the implementation of the asynchronous SSVEP-BCI system, leading to improved robustness and flexibility in human-machine interaction.

An examination of the status, contexts of anatomical body donation, and perspectives in China.

Body donation is a valuable resource in medical education, research, clinical diagnosis, and treatment. Consequently, donors are honored as "Silent Mentors" in Chinese medical schools. This article briefly reviews the history, current status, and strategies to promote body donation in China (excluding data from Hong Kong, Macao, and Taiwan regions) and discusses the problems encountered in body donation work in China. After establishing the People's Republic of China in 1949, the central government issued regulations on the use of dissected bodies. In 2001, the "Shanghai Regulations on Body Donation" were officially implemented and became China's first local legislative regulation on body donation. Subsequently, local legislative regulations and rules on body donation were issued in various regions to promote smooth and orderly body donation. There has been tremendous development in body donation in China for more than 40 years; however, the progress of this partial work has been uneven in various areas owing to the influence of traditional ethical concepts. It is, therefore, imperative to legislate body donations at a national level. Raising the public's scientific literacy and changing the traditional concept of funerals can create a positive social atmosphere for body donation, thus increasing the public's awareness and willingness to donate their bodies. Donating the body at the end of life contributes to life science and medical causes and is a noble act worthy of praise.

Genetically encoded bioorthogonal tryptophan decaging in living cells.

Tryptophan (Trp) plays a critical role in the regulation of protein structure, interactions and functions through its π system and indole N-H group. A generalizable method for blocking and rescuing Trp interactions would enable the gain-of-function manipulation of various Trp-containing proteins in vivo, but generating such a platform remains challenging. Here we develop a genetically encoded N1-vinyl-caged Trp capable of rapid and bioorthogonal decaging through an optimized inverse electron-demand Diels-Alder reaction, allowing site-specific activation of Trp on a protein of interest in living cells. This chemical activation of a genetically encoded caged-tryptophan (Trp-CAGE) strategy enables precise activation of the Trp of interest underlying diverse important molecular interactions. We demonstrate the utility of Trp-CAGE across various protein families, such as catalase-peroxidases and kinases, as translation initiators and posttranslational modification readers, allowing the modulation of epigenetic signalling in a temporally controlled manner. Coupled with computer-aided prediction, our strategy paves the way for bioorthogonal Trp activation on more than 28,000 candidate proteins within their native cellular settings.

Computational design and genetic incorporation of lipidation mimics in living cells.

Protein lipidation, which regulates numerous biological pathways and plays crucial roles in the pharmaceutical industry, is not encoded by the genetic code but synthesized post-translationally. In the present study, we report a computational approach for designing lipidation mimics that fully recapitulate the biochemical properties of natural lipidation in membrane association and albumin binding. Furthermore, we establish an engineered system for co-translational incorporation of these lipidation mimics into virtually any desired position of proteins in Escherichia coli and mammalian cells. We demonstrate the utility of these length-tunable lipidation mimics in diverse applications, including improving the half-life and activity of therapeutic proteins in living mice, anchoring functional proteins to membrane by substituting natural lipidation, functionally characterizing proteins carrying different lengths of lipidation and determining the plasma membrane-binding capacity of a given compound. Our strategy enables gain-of-function studies of lipidation in hundreds of proteins and facilitates the creation of superior therapeutic candidates.

New Strategies for Probing the Biological Functions of Protein Post-translational Modifications in Mammalian Cells with Genetic Code Expansion.

Protein post-translational modification (PTM) is a major mechanism for functional diversification of the human genome and plays a crucial role in almost every aspect of cellular processes, and the dysregulation of the protein PTM network has been associated with a variety of human diseases. Using high-resolution mass spectrometry, protein PTMs can be efficiently discovered and profiled under various biological and physiological conditions. However, it is often challenging to address the biological function of PTMs with biochemical and mutagenesis-based approaches. Specifically, this field lacks methods that allow gain-of-function studies of protein PTMs to understand their functional consequences in living cells. In this context, the genetic code expansion (GCE) strategy has made tremendous progress in the direct installation of PTMs and their analogs in the form of noncanonical amino acids (ncAAs) for gain-of-function investigations.In addition to studying the biological functions of known protein PTMs, the discovery of new protein PTMs is even more challenging due to the lack of chemical information for designing specific enrichment methods. Genetically encoded ncAAs in the proteome can be used as specific baits to enrich and subsequently identify new PTMs by mass spectrometry.In this Account, we discuss recent developments in the investigation of the biological functions of protein PTMs and the discovery of protein PTMs using new GCE strategies. First, we leveraged a chimeric design to construct several broadly orthogonal translation systems (OTSs). These broad OTSs can be engineered to efficiently incorporate different ncAAs in both E. coli and mammalian cells. With these broad OTSs, we accomplish the following: (1) We develop a computer-aided strategy for the design and genetic incorporation of length-tunable lipidation mimics. These lipidation mimics can fully recapitulate the biochemical properties of natural lipidation in membrane association for probing its biological functions on signaling proteins and in albumin binding for designing long-acting protein drugs. (2) We demonstrate that the binding affinity between histone methylations and their corresponding readers can be substantially increased with genetically encoded electron-rich Trp derivatives. These engineered affinity-enhanced readers can be applied to enrich, image, and profile the interactome of chromatin methylations. (3) We report the identification and verification of a novel type of protein PTM, aminoacylated lysine ubiquitination, using genetically encoded PTM ncAAs as chemical probes. This approach provides a general strategy for the identification of unknown PTMs by increasing the abundance of PTM bait probes.

Rapid Construction of Caffeic Acid/p-Phenylenediamine Antifouling Hydrophilic Coating on a PVDF Membrane for Emulsion Separation.

The current methods of constructing modification strategies for hydrophilic membranes are time-consuming, complex in operation, and poor in universality, which limit their application on membranes. In this work, inspired by the adhesion properties and versatility of caffeic acid (CA) and p-phenylenediamine (PPDA), a simple, rapid, and universal method was designed for the separation of oil-in-water emulsion by preparing a stable hydrophilic coating separation membrane. The preparation time of the membrane was shortened to 40 min. The developed PVDF-PCA/PPDA membrane showed superhydrophilic and underwater superoleophobic properties. When applied to petroleum ether-in-water emulsion, isooctane-in-water emulsion, and dodecane-in-water emulsion separation, the oil rejection was more than 99.0%. In the circulating separation of 10 g/L soybean oil-in-water emulsion, the oil rejection was more than 99.3%, and the highest flux was 1036 L·m-2·h-1. The prepared PVDF-PCA/PPDA membrane performed well in the separation test of oily wastewater. The proposed strategy is simple and rapid; it may become a universal method for preparing membranes with super strong antifouling properties against viscous oil and accelerate the research progress of membrane separation of oil-in-water emulsions.

Manipulating Cation-π Interactions of Reader Proteins in Living Cells with Genetic Code Expansion.

Despite tremendous success in understanding the chemical nature and the importance of cation-π interactions in a range of biological processes, particularly in epigenetic regulation, the design and synthesis of stronger cation-π interactions in living cells remain largely elusive. Here, we design several electron-rich Trp derivatives and incorporate them into histone methylation reader domains to enhance the affinity of the reader domains for histone methylation marks via cation-π interactions in living cells. We show that this site-specific Trp replacement strategy is generally applicable for the engineering of high-affinity reader domains for the major histone H3 trimethylation marks, such as H3K4me3, H3K9me3, H3K27me3, and H3K36me3, with high specificity. Furthermore, we demonstrate that engineered reader domains can serve as powerful tools for the enrichment and imaging of histone methylation, as well as for capturing the protein interactome at chromatin marks in living cells. Therefore, our study paves the way for the design of enhanced cation-π interactions in reader proteins in living cells for various biological applications.

Genetic code expansion reveals aminoacylated lysine ubiquitination mediated by UBE2W.

Protein post-translational modification (PTM) regulates nearly every aspect of cellular processes in eukaryotes. However, the identification of new protein PTMs is very challenging. Here, using genetically encoded unnatural amino acids as chemical probes, we report the identification and validation of a previously unreported form of protein PTM, aminoacylated lysine ubiquitination, in which the modification occurs on the α-amine group of aminoacylated lysine. We identify more than 2,000 ubiquitination sites on all 20 aminoacylated lysines in two human cell lines. The modifications can mediate rapid protein degradation, complementing the canonical lysine ubiquitination-mediated proteome degradation. Furthermore, we demonstrate that the ubiquitin-conjugating enzyme UBE2W acts as a writer of aminoacylated lysine ubiquitination and facilitates the ubiquitination event on proteins. More broadly, the discovery and validation of aminoacylated lysine ubiquitination paves the way for the identification and verification of new protein PTMs with the genetic code expansion strategy.

UBIAD1 effectively alleviated myocardial ischemia reperfusion injury by activating SIRT1/PGC1α.

BACKGROUND AND OBJECTIVES: Myocardial ischemia-reperfusion injury (MIRI) threatens global health and lowers people's sense of happiness. Till now, the mechanism of MIRI has not been well-understood. Therefore, this study was designed to explore the role of UBIAD1 in MIRI as well as its detailed reaction mechanism. METHODS: The mRNA and protein expressions of UBIAD1 before or after transfection were measured using RT-qPCR and western blot. Western blot was also adopted to measure the expressions of signaling pathway-, mitochondrial damage- and apoptosis-related proteins. Moreover, mitochondrial membrane potential and ATP level were verified by JC-1 immunofluorescence and ATP kits, respectively. With the application of CCK-8, LDH and CK-MB assays, the cell viability, LDH and CK-MB levels were evaluated, respectively. In addition, the cell apoptosis was detected using TUNEL. Finally, the expressions of ROS, SOD, MDA and CAT were measured using DCFH-DA, SOD, MDA and CAT assays, respectively. RESULTS: In the present study, we found that UBIAD1 was downregulated in hypoxia-reoxygenation (H/R) -induced H9C2 cells and its upregulation could activate SIRT1/PGC1α signaling pathway. It was also found that UBIAD1 regulated mitochondrial membrane potential and ATP level via activating SIRT1/PGC1α signaling pathway. In addition, the injury of H/R-induced H9C2 cells could be relieved by UBIAD1 through the activation of SIRT1/PGC1α signaling pathway. Moreover, UBIAD1 exhibited inhibitory effects on apoptosis and oxidative stress of H/R-induced H9C2 cells through activating SIRT1/PGC1α signaling pathway. CONCLUSION: To sum up, UBIAD1 could alleviate apoptosis, oxidative stress and H9C2 cell injury by activating SIRT1/PGC1α, which laid experimental foundation for the clinical treatment of MIRI.

Safety-oriented planning of expressway truck service areas based on driver demand.

The rapid development of the economy has promoted the growth of freight transportation. The truck service areas on expressways, as the main places for truck drivers to rest, play an important role in ensuring the driving safety of trucks. If these service areas are constructed densely or provide a plentiful supply of parking areas, they are costly to construct. However, if the distance between two adjacent truck service areas is very large or the number of truck parking spaces in service areas is small, the supply will fail to meet the parking needs of truck drivers. In this situation, the continuous working time of truck drivers will be longer, and this is likely to cause driver fatigue and even traffic accidents. To address these issues, this paper established a non-linear optimization model for truck service area planning of expressways to optimize truck driving safety. An improved genetic algorithm is proposed to solve the model. A case study of a 215.5-kilometers-length section of the Guang-Kun expressway in China was used to demonstrate the effectiveness of the model and algorithm. As validated by this specific case, the proposed model and solution algorithm can provide an optimal plan for the layout of truck service areas that meet the parking needs of truck drivers while minimizing the service loss rate. The research results of this paper can contribute to the construction of truck service areas and the parking management of trucks on expressways.

Simultaneous sEMG Recognition of Gestures and Force Levels for Interaction With Prosthetic Hand.

The natural interaction between the prosthetic hand and the upper limb amputation patient is important and directly affects the rehabilitation effect and operation ability. Most previous studies only focused on the interaction of gestures but ignored the force levels. This paper proposes a simultaneous recognition method of gestures and forces for interaction with a prosthetic hand. The multitask classification algorithm based on a convolutional neural network (CNN) is designed to improve recognition efficiency and ensure recognition accuracy. The offline experimental results show that the algorithm proposed in this study outperforms other methods in both training speed and accuracy. To prove the effectiveness of the proposed method, a myoelectric prosthetic hand integrated with tactile sensors is developed, and surface electromyography (sEMG) datasets of healthy persons and amputees are built. The online experimental results show that the amputee can control the prosthetic hand to continuously make gestures under different force levels, and the effect of hand coordination on the hand perception of amputees is explored. The results show that gesture classification operation tasks with different force levels based on sEMG signals can be accurately recognized and comfortably interact with prosthetic hands in real time. It improves the amputees' operation ability and relieves their muscle fatigue.

New Far-Red and Near-Infrared Fluorescent Phycobiliproteins with Excellent Brightness and Photostability.

Far-red and near-infrared fluorescent proteins can be used as fluorescence biomarkers in the region of maximal transmission of most tissues and facilitate multiplexing. Recently, we reported the generation and properties of far-red and near-infrared fluorescent phycobiliproteins, termed BeiDou Fluorescent Proteins (BDFPs), which can covalently bind the more readily accessible biliverdin. Far-red BDFPs maximally fluoresce at ∼670 nm, while near-infrared BDFPs fluoresce at ∼710 nm. In this work, we molecularly evolved BDFPs as follows: (a) mutations L58Q, S68R and M81K of BDFPs, which can maximally enhance the effective brightness in vivo by 350 %; (b) minimization and monomerization of far-red BDFPs 2.1, 2.2, 2.3, and near-infrared BDFPs 2.4, 2.5 and 2.6. These newly developed BDFPs are remarkably brighter than the formerly reported far-red and near-infrared fluorescent proteins. Their advantages are demonstrated by biolabeling in mammalian cells using super-resolution microscopy.

Manipulating Cation-π Interactions with Genetically Encoded Tryptophan Derivatives.

Cation-π interactions are the major noncovalent interactions for molecular recognition and play a central role in a broad area of chemistry and biology. Despite tremendous success in understanding the origin and biological importance of cation-π interactions, the design and synthesis of stronger cation-π interactions remain elusive. Here, we report an approach that greatly increases the binding energy of cation-π interactions by replacing Trp in the aromatic box with an electron-rich Trp derivative using the genetic code expansion strategy. The binding affinity between histone H3K4me3 and its reader is increased more than eightfold using genetically encoded 6-methoxy-Trp. Furthermore, through a systematic engineering process, we construct an H3K4me3 Super-Reader with single-digit nM affinity for H3K4me3 detection and imaging. More broadly, this approach paves the way for manipulating cation-π interactions for a variety of applications.

Defective Bimetallic Selenides for Selective CO2 Electroreduction to CO.

CO2 electroreduction (CO2 RR) to CO is promising for the carbon cycle but still remains challenging. Au is regarded as the most selective catalyst for CO2 RR, but its high cost significantly hinders its industrial application. Herein, the bimetallic CuInSe2 is found to exhibit an Au-like catalytic feature: i) the interaction of Cu and In orbitals induces a moderate adsorption strength of CO2 RR intermediates and favors the reaction pathway; and ii) the hydrogen evolution is energetically unfavorable on CuInSe2 , as a surface reconstruction along with high energy change will occur after hydrogen adsorption. Furthermore, the Se vacancy is found to induce an electron redistribution, slightly tune the band structure, and optimize the CO2 RR route of bimetallic selenide. Consequently, the Se-defective CuInSe2 (V-CuInSe2 ) achieves a highly selective CO production ability that is comparable to noble metals in aqueous electrolyte, and the V-CuInSe2 cathode shows a satisfactory performance in an aqueous Zn-CO2 cell. This work demonstrates that designing cost-effective catalysts with noble-metal-like properties is an ideal strategy for developing efficient electrocatalysts. Moreover, the class of transition bimetallic selenides has shown promising prospects as active and cost-effective electrocatalysts owing to their unique structural, electronic, and catalytic properties.

Filter Bank Convolutional Neural Network for Short Time-Window Steady-State Visual Evoked Potential Classification.

Convolutional neural network (CNN) has been gradually applied to steady-state visual evoked potential (SSVEP) of the brain-computer interface (BCI). Frequency-domain features extracted by fast Fourier Transform (FFT) or time-domain signals are used as network input. In the frequency-domain diagram, the features at the short time-window are not obvious and the phase information of each electrode channel may be ignored as well. Hence we propose a time-domain-based CNN method (tCNN), using the time-domain signal as network input. And the filter bank tCNN (FB-tCNN) is further proposed to improve its performance in the short time-window. We compare FB-tCNN with the canonical correlation analysis (CCA) methods and other CNN methods in our dataset and public dataset. And FB-tCNN shows superior performance at the short time-window in the intra-individual test. At the 0.2 s time-window, the accuracy of our method reaches 88.36 ± 4.89 % in our dataset, 77.78 ± 2.16 % and 79.21 ± 1.80 % respectively in the two sessions of the public dataset, which is higher than other methods. The impacts of training-subject number and data length in inter-individual or cross-individual are studied. FB-tCNN shows the potential in implementing inter-individual BCI. Further analysis shows that the deep learning method is easier in terms of the implementation of the asynchronous BCI system than the training data-driven CCA. The code is available for reproducibility at https://github.com/DingWenl/FB-tCNN.