Immunogenicity prediction is a central topic in reverse vaccinology for finding candidate vaccines that can trigger protective immune responses. Existing approaches typically rely on highly compressed features and simple model architectures, leading to limited prediction accuracy and poor generalizability. To address these challenges, we introduce VenusVaccine, a novel deep learning solution with a dual attention mechanism that integrates pre-trained latent vector representations of protein sequences and structures. We also compile the most comprehensive immunogenicity dataset to date, encompassing over 7000 antigen sequences, structures, and immunogenicity labels from bacteria, viruses, and tumors. Extensive experiments demonstrate that VenusVaccine outperforms existing methods across a wide range of evaluation metrics. Furthermore, we establish a post-hoc validation protocol to assess the practical significance of deep learning models in tackling vaccine design challenges. Our work provides an effective tool for vaccine design and sets valuable benchmarks for future research. The implementation is at \url{https://github.com/songleee/VenusVaccine}.

The inertia level of microgrids decreases with the access of inverter-based distributed generations. Uncertainty of source loads in islanded microgrids can lead to frequency instability. Considering electric vehicles (EVs) as a typical regulable load, a charging station (CS) virtual synchronous generator (VSG) frequency control (FC) strategy that considers the user charging and grid FC demand is proposed. First, the control architecture of CS based on the VSG is designed. Second, considering the impact of VSG FC on EV user charging demand, the charging power response model considering the charging power FC cost is designed. Furthermore, multiobjective model predictive control algorithm is designed to improve the VSG control strategy, and the FC strategy is obtained. Simulation results show that the proposed method can provide inertial support and frequency regulation capability by the CS with VSG control strategy while satisfying EV charging demand. In addition, the FC cost can be decreased.

Drug-target binding affinity prediction is a fundamental task for drug discovery and has been studied for decades. Most methods follow the canonical paradigm that processes the inputs of the protein (target) and the ligand (drug) separately and then combines them together. In this study we demonstrate, surprisingly, that a model is able to achieve even superior performance without access to any protein-sequence-related information. Instead, a protein is characterized completely by the ligands that it interacts. Specifically, we treat different proteins separately, which are jointly trained in a multi-head manner, so as to learn a robust and universal representation of ligands that is generalizable across proteins. Empirical evidences show that the novel paradigm outperforms its competitive sequence-based counterpart, with the Mean Squared Error (MSE) of 0.4261 versus 0.7612 and the R-Square of 0.7984 versus 0.6570 compared with DeepAffinity. We also investigate the transfer learning scenario where unseen proteins are encountered after the initial training, and the cross-dataset evaluation for prospective studies. The results reveals the robustness of the proposed model in generalizing to unseen proteins as well as in predicting future data. Source codes and data are available at https://github.com/huzqatpku/SAM-DTA.

Rapid and selective ion transport is essential for the generation and regulation of electrical signaling pathways in living organisms. Here, we use molecular dynamics (MD) simulations with an applied membrane potential to investigate the ion flux of bacterial sodium channel NaVMs. 5.9 µs simulations with 500 mM NaCl suggest different mechanisms for inward and outward flux. The predicted inward conductance rate of ∼27±3 pS, agrees with experiment. The estimated outward conductance rate is 15±3 pS, which is considerably lower. Comparing inward and outward flux, the mean ion dwell time in the selectivity filter (SF) is prolonged from 13.5±0.6 ns to 20.1±1.1 ns. Analysis of the Na+ distribution revealed distinct patterns for influx and efflux events. In 32.0±5.9% of the simulation time, the E53 side chains adopted a flipped conformation during outward conduction, whereas this conformational change was rarely observed (2.7±0.5%) during influx. Further, simulations with dihedral restraints revealed that influx is less affected by the E53 conformational flexibility. In contrast, during outward conduction, our simulations indicate that the flipped E53 conformation provides direct coordination for Na+. The free energy profile (potential of mean force calculations) indicates that this conformational change lowers the putative barriers between sites SCEN and SHFS during outward conduction. We hypothesize that during an action potential, the increased Na+ outward transition propensities at depolarizing potentials might increase the probability of E53 conformational changes in the SF. Subsequently, this might be a first step towards initiating slow inactivation.