In recent years, the application of artificial intelligence (AI) in the field of biology has witnessed remarkable advancements. Among these, the most notable achievements have emerged in the domain of protein structure prediction and design, with AlphaFold and related innovations earning the 2024 Nobel Prize in Chemistry. These breakthroughs have transformed our ability to understand protein folding and molecular interactions, marking a pivotal milestone in computational biology. Looking ahead, it is foreseeable that the accurate prediction of various physicochemical properties of proteins—beyond static structure—will become the next critical frontier in this rapidly evolving field. One of the most important protein properties is thermodynamic stability, which refers to a protein’s ability to maintain its native conformation under physiological or stress conditions. Accurate prediction of protein stability, especially upon single-point mutations, plays a vital role in numerous scientific and industrial domains. These include understanding the molecular basis of disease, rational drug design, development of therapeutic proteins, design of more robust industrial enzymes, and engineering of biosensors. Consequently, the ability to reliably forecast the stability changes caused by mutations has broad and transformative implications across biomedical and biotechnological applications. Historically, protein stability was assessed via experimental methods such as differential scanning calorimetry (DSC) and circular dichroism (CD), which, while precise, are time-consuming and resource-intensive. This prompted the development of computational approaches, including empirical energy functions and physics-based simulations. However, these traditional models often fall short in capturing the complex, high-dimensional nature of protein conformational landscapes and mutational effects. Recent advances in machine learning (ML) have significantly improved predictive performance in this area. Early ML models used handcrafted features derived from sequence and structure, whereas modern deep learning models leverage massive datasets and learn representations directly from data. Deep neural networks (DNNs), graph neural networks (GNNs), and attention-based architectures such as transformers have shown particular promise. GNNs, in particular, excel at modeling spatial and topological relationships in molecular structures, making them well-suited for protein modeling tasks. Furthermore, attention mechanisms enable models to dynamically weigh the contribution of specific residues or regions, capturing long-range interactions and allosteric effects. Nevertheless, several key challenges remain. These include the imbalance and scarcity of high-quality experimental datasets, particularly for rare or functionally significant mutations, which can lead to biased or overfitted models. Additionally, the inherently dynamic nature of proteins—their conformational flexibility and context-dependent behavior—is difficult to encode in static structural representations. Current models often rely on a single structure or average conformation, which may overlook important aspects of stability modulation. Efforts are ongoing to incorporate multi-conformational ensembles, molecular dynamics simulations, and physics-informed learning frameworks into predictive models. This paper presents a comprehensive review of the evolution of protein thermodynamic stability prediction techniques, with emphasis on the recent progress enabled by machine learning. It highlights representative datasets, modeling strategies, evaluation benchmarks, and the integration of structural and biochemical features. The aim is to provide researchers with a structured and up-to-date reference, guiding the development of more robust, generalizable, and interpretable models for predicting protein stability changes upon mutation. As the field moves forward, the synergy between data-driven AI methods and domain-specific biological knowledge will be key to unlocking deeper understanding and broader applications of protein engineering.

Alzheimer’s disease (AD) is a neurodegenerative disorder characterized by progressive cognitive decline, functional impairment, and neuropsychiatric symptoms. It represents the most prevalent form of dementia among the elderly population. Accumulating evidence indicates that oxidative stress plays a pivotal role in the pathogenesis of AD. Notably, elevated levels of oxidative stress have been observed in the brains of AD patients, where excessive reactive oxygen species (ROS) can cause extensive damage to lipids, proteins, and DNA, ultimately compromising neuronal structure and function. Amyloid β‑protein (Aβ) has been shown to induce mitochondrial dysfunction and calcium overload, thereby promoting the generation of ROS. This, in turn, exacerbates Aβ aggregation and enhances tau phosphorylation, leading to the formation of two pathological features of AD: extracellular Aβ plaque deposition and intracellular neurofibrillary tangles (NFTs). These events ultimately culminate in neuronal death, forming a vicious cycle. The interplay between oxidative stress and these pathological processes constitutes a core link in the pathogenesis of AD. The signaling pathways mediating oxidative stress in AD include Nrf2, RCAN1, PP2A, CREB, Notch1, NF‑κB, ApoE, and ferroptosis. Nrf2 signaling pathway serves as a key regulator of cellular redox homeostasis, exerts important antioxidant capacity and protective effects in AD. RCAN1 signaling pathway, as a calcineurin inhibitor, and modulates AD progression through multiple mechanisms. PP2A signaling pathway is involved in regulating tau phosphorylation and neuroinflammation processes. CREB signaling pathway contributes to neuroplasticity and memory formation; activation of CREB improves cognitive function and reduce oxidative stress. Notch1 signaling pathway regulates neuronal development and memory, participates in modulation of Aβ production, and interacts with Nrf2 toco-regulate antioxidant activity. NF‑κB signaling pathway governs immune and inflammatory responses; sustained activation of this pathway forms “inflammatory memory”, thereby exacerbating AD pathology. ApoE signaling pathway is associated with lipid metabolism; among its isoforms, ApoE-ε4 significantly increases the risk of AD, leading to elevated oxidative stress, abnormal lipid metabolism, and neuroinflammation. The ferroptosis signaling pathway is driven by iron-dependent lipid peroxidation, and the subsequent release of lipid peroxidation products and ROS exacerbate oxidative stress and neuronal damage. These interconnected pathways form a complex regulatory network that regulates the progression of AD through oxidative stress and related pathological cascades. In terms of therapeutic strategies targeting oxidative stress, among the drugs currently used in clinical practice for AD treatment, memantine and donepezil demonstrate significant therapeutic efficacy and can improve the level of oxidative stress in AD patients. Some compounds with antioxidant effects (such asα-lipoic acid and melatonin) have shown certain potential in AD treatment research and can be used as dietary supplements to ameliorate AD symptoms. In addition, non-drug interventions such as calorie restriction and exercise have been proven to exerted neuroprotective effects and have a positive effect on the treatment of AD. By comprehensively utilizing the therapeutic characteristics of different signaling pathways, it is expected that more comprehensive multi-target combination therapy regimens and combined nanomolecular delivery systems will be developed in the future to bypass the blood-brain barrier, providing more effective therapeutic strategies for AD.

Alzheimer's disease (AD), a prototypical neurodegenerative disorder, encompasses multifaceted pathological processes. As pivotal cellular structures within the central nervous system, ion channels play critical roles in regulating neuronal excitability, synaptic transmission, and neurotransmitter release. Extensive research has revealed significant alterations in the expression and function of ion channels in AD, implicating an important role of ion channels in the pathogenesis of abnormal Aβ deposition, neuroinflammation, oxidative stress, and disruptions in calcium homeostasis and neural network functionality. This review systematically summarizes the crucial roles and underlying mechanisms of ion channels in the onset and progression of AD, highlighting how these channel abnormalities contribute to AD pathophysiology. We also discuss the therapeutic potential of ion channel modulation in AD treatment, emphasizing the importance of addressing multifactorial nature and heterogeneity of AD. The development of multi-target drugs and precision therapies is proposed as a future direction of scientific research.
Alzheimer Disease/therapy* ; Humans ; Ion Channels/physiology* ; Oxidative Stress ; Animals ; Amyloid beta-Peptides/metabolism* ; Synaptic Transmission ; Calcium/metabolism*

This study aimed to characterize and identify the non-volatile components in aqueous and ethanolic extracts of the stems and leaves of Rhododendron tomentosum by using sensitive and efficient ultra-performance liquid chromatography-quadrupole-time of flight mass spectrometry(UHPLC-Q-TOF-MS) combined with a self-built information database. By comparing with reference compounds, analyzing fragment ion information, searching relevant literature, and using a self-built information database, 118 compounds were identified from the aqueous and ethanolic extracts of R. tomentosum, including 35 flavonoid glycosides, 15 phenolic glycosides, 12 flavonoids, 7 phenolic acids, 7 phenylethanol glycosides, 6 tannins, 6 phospholipids, 5 coumarins, 5 monoterpene glycosides, 6 triterpenes, 3 fatty acids, and 11 other types of compounds. Among them, 102 compounds were reported in R. tomentosum for the first time, and 36 compounds were identified by comparing them with reference compounds. The chemical components in the ethanolic and aqueous extracts of R. tomentosum leaves and stems showed slight differences, with 84 common chemical components accounting for 71.2% of the total 118 compounds. This study systematically characterized and identified the non-volatile chemical components in the ethanolic and aqueous extracts of R. tomentosum for the first time. The findings provide a reference for active ingredient research, quality control, and product development of R. tomentosum.
Rhododendron/chemistry* ; Chromatography, High Pressure Liquid/methods* ; Drugs, Chinese Herbal/chemistry* ; Mass Spectrometry/methods* ; Plant Leaves/chemistry*

Transformer models have emerged as pivotal tools within the realm of drug discovery, distinguished by their unique architectural features and exceptional performance in managing intricate data landscapes. Leveraging the innate capabilities of transformer architectures to comprehend intricate hierarchical dependencies inherent in sequential data, these models showcase remarkable efficacy across various tasks, including new drug design and drug target identification. The adaptability of pre-trained transformer-based models renders them indispensable assets for driving data-centric advancements in drug discovery, chemistry, and biology, furnishing a robust framework that expedites innovation and discovery within these domains. Beyond their technical prowess, the success of transformer-based models in drug discovery, chemistry, and biology extends to their interdisciplinary potential, seamlessly combining biological, physical, chemical, and pharmacological insights to bridge gaps across diverse disciplines. This integrative approach not only enhances the depth and breadth of research endeavors but also fosters synergistic collaborations and exchange of ideas among disparate fields. In our review, we elucidate the myriad applications of transformers in drug discovery, as well as chemistry and biology, spanning from protein design and protein engineering, to molecular dynamics (MD), drug target identification, transformer-enabled drug virtual screening (VS), drug lead optimization, drug addiction, small data set challenges, chemical and biological image analysis, chemical language understanding, and single cell data. Finally, we conclude the survey by deliberating on promising trends in transformer models within the context of drug discovery and other sciences.

Objective To investigate the value of magnetic resonance imaging(MRI)T2-mapping in evaluating the activity of Graves ophthalmopathy(GO).Methods A total of 64 patients with GO in the Department of Endocrinology,the First Affiliated Hospital of Chongqing Medical University from July 2019 to January 2021 were collected.Simple random grouping was performed by computer,with 49 cases as observation subjects,and 15 patients for diagnostic test.According to clinical activity score(CAS),49 GO patients were divided into active group(CAS≥3 points,48 eyes)and inactive group(CAS<3 points,50 eyes).Normal control group(NC group)included 31 patients(62 eyes).All subjects underwent 3.0T orbital MRI T2-mapping.Measuring the T2 relaxation time(T2RT)of superior rectus,inferior rectus,medial rectus,and lateral rectus on five layers behind the eyeball on T2-mapping coronal images,and select the maximum value of T2RT in the five layers for each extraocular muscle to represent the T2RT of this extraocular muscle.Finally,select the maximum T2RT values of the four extraocular muscles,expressed as extraocular muscle maximum T2RT.Compare the differences of the above 5 indicators(superior rectus T2RT,inferior rectus T2RT,medial rectus T2RT,lateral rectus T2RT,extraocular muscle maximum T2RT)between active group,inactive group and NC group.ROC curve was used to analyze the diagnostic value of the above 5 indicators for GO activity assessment,and the diagnostic threshold was obtained.Then,another 15 GO patients were performed for diagnostic tests evaluation to determine the indicators of high diagnostic efficacy and the threshold of diagnostic activity.Results The T2RT of all extraocular muscles in active group were significantly higher than those in inactive group and NC group,the difference was statistically significant(P<0.001).The threshold value of the five indicators were obtained by ROC curve analysis.The maximum T2RT cut-off values of superior rectus muscle,inferior rectus muscle,medial rectus muscle,lateral rectus muscle and extraocular muscles for judging activity were 80.200 ms,97.045 ms,94.355 ms,85.750 ms and 101.385 ms respectively.Another 15 GO patients were performed for diagnostic tests,the indexes with relatively high sensitivity,specificity,positive predictive value and negative predictive value were inferior rectus T2RT and extraocular muscle maximum T2RT,the cut-off values of GO activity were 97.045 ms and 101.385 ms,respectively;the sensitivity were 91.7％and 93.8％,respectively;the specificity all were 80.0％.Conclusions MRI T2-mapping sequence has a good value in assessment of GO activity.The inferior rectus T2RT and extraocular muscle maximum T2RT can be choosed to evaluate the activity of GO.

Objective To develop a multi-modal deep learning method for automatic classification of immune-mediated glomerular diseases based on images of optical microscopy(OM),immunofluorescence microscopy(IM),and transmission electron microscopy(TEM).Methods We retrospectively collected the pathological images from 273 patients and constructed a multi-modal multi-instance model for classification of 3 immune-mediated glomerular diseases,namely immunoglobulin A nephropathy(IgAN),membranous nephropathy(MN),and lupus nephritis(LN).This model adopts an instance-level multi-instance learning(I-MIL)method to select the TEM images for multi-modal feature fusion with the OM images and IM images of the same patient.By comparing this model with unimodal and bimodal models,we explored different combinations of the 3 modalities and the optimal methods for modal feature fusion.Results The multi-modal multi-instance model combining OM,IM,and TEM images had a disease classification accuracy of(88.34±2.12)%,superior to that of the optimal unimodal model[(87.08±4.25)%]and that of the optimal bimodal model[(87.92±3.06)%].Conclusion This multi-modal multi-instance model based on OM,IM,and TEM images can achieve automatic classification of immune-mediated glomerular diseases with a good classification accuracy.

In infants with severe bronchopulmonary dysplasia(sBPD),severe pulmonary lobar emphysema may occur as a complication,contributing to significant impairment in ventilation.Clinical management of these infants is extremely challenging and some may require lobectomy to improve ventilation.However,prior to the lobectomy,it is very difficult to assess whether the remaining lung parenchyma would be able to sustain adequate ventilation postoperatively.In addition,preoperative planning and perioperative management are also quite challenging in these patients.This paper reports the utility of selective bronchial occlusion in assessing the safety and efficacy of lobectomy in a case of sBPD complicated by severe right upper lobar emphysema.Since infants with sBPD already have poor lung development and significant lung injury,lobectomy should be viewed as a non-traditional therapy and be carried out with extreme caution.Selective bronchial occlusion test can be an effective tool in assessing the risks and benefits of lobectomy in cases with sBPD and lobar emphysema.However,given the technical difficulty,successful application of this technique requires close collaboration of an experienced interdisciplinary team.

With the improvement of China's socioeconomic status,the issue of aging has become increasingly prominent,making cerebral infarction a common disease among the elderly.In recent years,research on cerebral infarction has gradually deepened,shifting focus from merely protecting and repairing neurons to emphasizing the complex interplay between inflammatory response and autophagy in the brain vascular unit,covering various aspects such as the blood-brain barrier,astrocytes,microglia,and autophagy.This shift in research direction has provided us with a profound understanding of the mechanisms underlying cerebral infarction,offering strong support for innovative future treatment strategies.In this review,we delved into the importance of the interplay between inflammatory response and autophagy in the pathogenesis of cerebral infarction,emphasized the intricate interactions among these biological components,which might lay the groundwork for more effective managements and treatments of cerebral infarction.By comprehensively reviewing existing literatures,we proposed future research directions,aiming to provide more scientific and systematic guidance for the clinical management and treatment of cerebral infarction.

Objective To develop a multi-modal deep learning method for automatic classification of immune-mediated glomerular diseases based on images of optical microscopy(OM),immunofluorescence microscopy(IM),and transmission electron microscopy(TEM).Methods We retrospectively collected the pathological images from 273 patients and constructed a multi-modal multi-instance model for classification of 3 immune-mediated glomerular diseases,namely immunoglobulin A nephropathy(IgAN),membranous nephropathy(MN),and lupus nephritis(LN).This model adopts an instance-level multi-instance learning(I-MIL)method to select the TEM images for multi-modal feature fusion with the OM images and IM images of the same patient.By comparing this model with unimodal and bimodal models,we explored different combinations of the 3 modalities and the optimal methods for modal feature fusion.Results The multi-modal multi-instance model combining OM,IM,and TEM images had a disease classification accuracy of(88.34±2.12)%,superior to that of the optimal unimodal model[(87.08±4.25)%]and that of the optimal bimodal model[(87.92±3.06)%].Conclusion This multi-modal multi-instance model based on OM,IM,and TEM images can achieve automatic classification of immune-mediated glomerular diseases with a good classification accuracy.