Large-volume fully automated cell reconstruction generates a cell atlas of plant tissues.

The geometric shape and arrangement of individual cells play a role in shaping organ functions. However, analyzing multicellular features and exploring their connectomes in centimeter-scale plant organs remain challenging. Here, we established a set of frameworks named Large-Volume Fully Automated Cell Reconstruction (LVACR), enabling the exploration of three-dimensional (3D) cytological features and cellular connectivity in plant tissues. Through benchmark testing, our framework demonstrated superior efficiency in cell segmentation and aggregation, successfully addressing the inherent challenges posed by light sheet fluorescence microscopy (LSFM) imaging. Using LVACR, we successfully established a cell atlas of different plant tissues. Cellular morphology analysis revealed differences of cell clusters and shapes in between different poplar (P. simonii Carr. and P. canadensis Moench.) seeds, whereas topological analysis revealed that they maintained conserved cellular connectivity. Furthermore, LVACR spatiotemporally demonstrated an initial burst of cell proliferation, accompanied by morphological transformations at an early stage in developing the shoot apical meristem. During subsequent development, cell differentiation produced anisotropic features, thereby resulting in various cell shapes. Overall, our findings provided valuable insights into the precise spatial arrangement and cellular behavior of multicellular organisms, thus enhancing our understanding of the complex processes underlying plant growth and differentiation.

3-D bioprinted human-derived skin organoids accelerate full-thickness skin defects repair.

The healing of large skin defects remains a significant challenge in clinical settings. The lack of epidermal sources, such as autologous skin grafting, limits full-thickness skin defect repair and leads to excessive scar formation. Skin organoids have the potential to generate a complete skin layer, supporting in-situ skin regeneration in the defect area. In this study, skin organoid spheres, created with human keratinocytes, fibroblasts, and endothelial cells, showed a specific structure with a stromal core surrounded by surface keratinocytes. We selected an appropriate bioink and innovatively combined an extrusion-based bioprinting technique with dual-photo source cross-linking technology to ensure the overall mechanical properties of the 3D bioprinted skin organoid. Moreover, the 3D bioprinted skin organoid was customized to match the size and shape of the wound site, facilitating convenient implantation. When applied to full-thickness skin defects in immunodeficient mice, the 3D bioprinted human-derived skin organoid significantly accelerated wound healing through in-situ regeneration, epithelialization, vascularization, and inhibition of excessive inflammation. The combination of skin organoid and 3D bioprinting technology can overcome the limitations of current skin substitutes, offering a novel treatment strategy to address large-area skin defects.

Advanced glycation end products mediate biomineralization disorder in diabetic bone disease.

Patients with diabetes often experience fragile fractures despite normal or higher bone mineral density (BMD), a phenomenon termed the diabetic bone paradox (DBP). The pathogenesis and therapeutics opinions for diabetic bone disease (DBD) are not fully explored. In this study, we utilize two preclinical diabetic models, the leptin receptor-deficient db/db mice (DB) mouse model and the streptozotocin-induced diabetes (STZ) mouse model. These models demonstrate higher BMD and lower mechanical strength, mirroring clinical observations in diabetic patients. Advanced glycation end products (AGEs) accumulate in diabetic bones, causing higher non-enzymatic crosslinking within collagen fibrils. This inhibits intrafibrillar mineralization and leads to disordered mineral deposition on collagen fibrils, ultimately reducing bone strength. Guanidines, inhibiting AGE formation, significantly improve the microstructure and biomechanical strength of diabetic bone and enhance bone fracture healing. Therefore, targeting AGEs may offer a strategy to regulate bone mineralization and microstructure, potentially preventing the onset of DBD.

Multifunctional hydrogel-based engineered extracellular vesicles delivery for complicated wound healing.

The utilization of extracellular vesicles (EVs) in wound healing has been well-documented. However, the direct administration of free EVs via subcutaneous injection at wound sites may result in the rapid dissipation of bioactive components and diminished therapeutic efficacy. Functionalized hydrogels provide effective protection, as well as ensure the sustained release and bioactivity of EVs during the wound healing process, making them an ideal candidate material for delivering EVs. In this review, we introduce the mechanisms by which EVs accelerate wound healing, and then elaborate on the construction strategies for engineered EVs. Subsequently, we discuss the synthesis strategies and application of hydrogels as delivery systems for the sustained release of EVs to enhance complicated wound healing. Furthermore, in the face of complicated wounds, functionalized hydrogels with specific wound microenvironment regulation capabilities, such as antimicrobial, anti-inflammatory, and immune regulation, used for loading engineered EVs, provide potential approaches to addressing these healing challenges. Ultimately, we deliberate on potential future trajectories and outlooks, offering a fresh viewpoint on the advancement of artificial intelligence (AI)-energized materials and 3D bio-printed multifunctional hydrogel-based engineered EVs delivery dressings for biomedical applications.

Light-activated nanoclusters with tunable ROS for wound infection treatment.

Infected wounds pose a significant clinical challenge due to bacterial resistance, recurrent infections, and impaired healing. Reactive oxygen species (ROS)-based strategies have shown promise in eradicating bacterial infections. However, the excess ROS in the infection site after treatments may cause irreversible damage to healthy tissues. To address this issue, we developed bovine serum albumin-iridium oxide nanoclusters (BSA-IrOx NCs) which enable photo-regulated ROS generation and scavenging using near infrared (NIR) laser. Upon NIR laser irradiation, BSA-IrOx NCs exhibit enhanced photodynamic therapy, destroying biofilms and killing bacteria. When the NIR laser is off, the nanoclusters' antioxidant enzyme-like activities prevent inflammation and repair damaged tissue through ROS clearance. Transcriptomic and metabolomic analyses revealed that BSA-IrOx NCs inhibit bacterial nitric oxide synthase, blocking bacterial growth and biofilm formation. Furthermore, the nanoclusters repair impaired skin by strengthening cell junctions and reducing mitochondrial damage in a fibroblast model. In vivo studies using rat infected wound models confirmed the efficacy of BSA-IrOx NCs. This study presents a promising strategy for treating biofilm-induced infected wounds by regulating the ROS microenvironment, addressing the challenges associated with current ROS-based antibacterial approaches.

Nanoengineered 3D-printing scaffolds prepared by metal-coordination self-assembly for hyperthermia-catalytic osteosarcoma therapy and bone regeneration.

The integration of functional nanomaterials with tissue engineering scaffolds has emerged as a promising solution for simultaneously treating malignant bone tumors and repairing resected bone defects. However, achieving a uniform bioactive interface on 3D-printing polymer scaffolds with minimized microstructural heterogeneity remains a challenge. In this study, we report a facile metal-coordination self-assembly strategy for the surface engineering of 3D-printed polycaprolactone (PCL) scaffolds with nanostructured two-dimensional conjugated metal-organic frameworks (cMOFs) consisting of Cu ions and 2,3,6,7,10,11-hexahydroxytriphenylene (HHTP). A tunable thickness of Cu-HHTP cMOF on PCL scaffolds was achieved via the alternative deposition of metal ions and HHTP. The resulting composite PCL@Cu-HHTP scaffolds not only demonstrated potent photothermal conversion capability for efficient OS ablation but also promoted the bone repair process by virtue of their cell-friendly hydrophilic interfaces. Therefore, the cMOF-engineered dual-functional 3D-printing scaffolds show promising potential for treating bone tumors by offering sequential anti-tumor effects and bone regeneration capabilities. This work also presents a new avenue for the interface engineering of bioactive scaffolds to meet multifaceted demands in osteosarcoma-related bone defects.

KinomeMETA: a web platform for kinome-wide polypharmacology profiling with meta-learning.

Kinase-targeted inhibitors hold promise for new therapeutic options, with multi-target inhibitors offering the potential for broader efficacy while minimizing polypharmacology risks. However, comprehensive experimental profiling of kinome-wide activity is expensive, and existing computational approaches often lack scalability or accuracy for understudied kinases. We introduce KinomeMETA, an artificial intelligence (AI)-powered web platform that significantly expands the predictive range with scalability for predicting the polypharmacological effects of small molecules across the kinome. By leveraging a novel meta-learning algorithm, KinomeMETA efficiently utilizes sparse activity data, enabling rapid generalization to new kinase tasks even with limited information. This significantly expands the repertoire of accurately predictable kinases to 661 wild-type and clinically-relevant mutant kinases, far exceeding existing methods. Additionally, KinomeMETA empowers users to customize models with their proprietary data for specific research needs. Case studies demonstrate its ability to discover new active compounds by quickly adapting to small dataset. Overall, KinomeMETA offers enhanced kinome virtual profiling capabilities and is positioned as a powerful tool for developing new kinase inhibitors and advancing kinase research. The KinomeMETA server is freely accessible without registration at https://kinomemeta.alphama.com.cn/.

Variation pattern in the macromolecular (cellulose, hemicelluloses, lignin) composition of cell walls in Pinus tabulaeformis tree trunks at different ages as revealed using multiple techniques.

The plant cell wall is a complex, heterogeneous structure primarily composed of cellulose, hemicelluloses, and lignin. Exploring the variations in these three macromolecules over time is crucial for understanding wood formation to enhance chemical processing and utilization. Here, we comprehensively analyzed the chemical composition of cell walls in the trunks of Pinus tabulaeformis using multiple techniques. In situ analysis showed that macromolecules accumulated gradually in the cell wall as the plant aged, and the distribution pattern of lignin was opposite that of polysaccharides, and both showed heterogenous distribution patterns. In addition, gel permeation chromatography (GPC) results revealed that the molecular weights of hemicelluloses decreased while that of lignin increased with age. Two-dimensional heteronuclear single quantum coherence nuclear magnetic resonance (2D-HSQC NMR) analysis indicated that hemicelluloses mainly comprised galactoglucomannan and arabinoglucuronoxylan, and the lignin types were mainly comprised guaiacyl (G) and p-hydroxyphenyl (H) units with three main linkage types: ß-O-4, ß-ß, and ß-5. Furthermore, the C-O bond (ß-O-4) signals of lignin decreased while the C-C bonds (ß-ß and ß-5) signals increased over time. Taken together, these findings shed light on wood formation in P. tabulaeformis and lay the foundation for enhancing the processing and use of wood and timber products.

Three-dimensional reconstruction and multiomics analysis reveal a unique pattern of embryogenesis in Ginkgo biloba.

Ginkgo (Ginkgo biloba L.) is one of the earliest extant species in seed plant phylogeny. Embryo development patterns can provide fundamental evidence for the origin, evolution, and adaptation of seeds. However, the architectural and morphological dynamics during embryogenesis in G. biloba remain elusive. Herein, we obtained over 2,200 visual slices from 3 stages of embryo development using micro-computed tomography imaging with improved staining methods. Based on 3-dimensional (3D) spatiotemporal pattern analysis, we found that a shoot apical meristem with 7 highly differentiated leaf primordia, including apical and axillary leaf buds, is present in mature Ginkgo embryos. 3D rendering from the front, top, and side views showed 2 separate transport systems of tracheids located in the hypocotyl and cotyledon, representing a unique pattern of embryogenesis. Furthermore, the morphological dynamic analysis of secretory cavities indicated their strong association with cotyledons during development. In addition, we identified genes GbLBD25a (lateral organ boundaries domain 25a), GbCESA2a (cellulose synthase 2a), GbMYB74c (myeloblastosis 74c), GbPIN2 (PIN-FORMED 2) associated with vascular development regulation, and GbWRKY1 (WRKYGOK 1), GbbHLH12a (basic helix-loop-helix 12a), and GbJAZ4 (jasmonate zim-domain 4) potentially involved in the formation of secretory cavities. Moreover, we found that flavonoid accumulation in mature embryos could enhance postgerminative growth and seedling establishment in harsh environments. Our 3D spatial reconstruction technique combined with multiomics analysis opens avenues for investigating developmental architecture and molecular mechanisms during embryogenesis and lays the foundation for evolutionary studies of embryo development and maturation.

Boosting cartilage repair with silk fibroin-DNA hydrogel-based cartilage organoid precursor.

Osteoarthritis (OA), a common degenerative disease, is characterized by high disability and imposes substantial economic impacts on individuals and society. Current clinical treatments remain inadequate for effectively managing OA. Organoids, miniature 3D tissue structures from directed differentiation of stem or progenitor cells, mimic native organ structures and functions. They are useful for drug testing and serve as active grafts for organ repair. However, organoid construction requires extracellular matrix-like 3D scaffolds for cellular growth. Hydrogel microspheres, with tunable physical and chemical properties, show promise in cartilage tissue engineering by replicating the natural microenvironment. Building on prior work on SF-DNA dual-network hydrogels for cartilage regeneration, we developed a novel RGD-SF-DNA hydrogel microsphere (RSD-MS) via a microfluidic system by integrating photopolymerization with self-assembly techniques and then modified with Pep-RGDfKA. The RSD-MSs exhibited uniform size, porous surface, and optimal swelling and degradation properties. In vitro studies demonstrated that RSD-MSs enhanced bone marrow mesenchymal stem cells (BMSCs) proliferation, adhesion, and chondrogenic differentiation. Transcriptomic analysis showed RSD-MSs induced chondrogenesis mainly through integrin-mediated adhesion pathways and glycosaminoglycan biosynthesis. Moreover, in vivo studies showed that seeding BMSCs onto RSD-MSs to create cartilage organoid precursors (COPs) significantly enhanced cartilage regeneration. In conclusion, RSD-MS was an ideal candidate for the construction and long-term cultivation of cartilage organoids, offering an innovative strategy and material choice for cartilage regeneration and tissue engineering.

ROS-scavenging hydrogel as protective carrier to regulate stem cells activity and promote osteointegration of 3D printed porous titanium prosthesis in osteoporosis.

Stem cell-based therapy has drawn attention as an alternative option for promoting prosthetic osteointegration in osteoporosis by virtue of its unique characteristics. However, estrogen deficiency is the main mechanism of postmenopausal osteoporosis. Estrogen, as an effective antioxidant, deficienncy also results in the accumulation of reactive oxygen species (ROS) in the body, affecting the osteogenic differentiation of stem cells and the bone formation i osteoporosis. In this study, we prepared a ROS-scavenging hydrogel by crosslinking of epigallocatechin-3-gallate (EGCG), 3-acrylamido phenylboronic acid (APBA) and acrylamide. The engineered hydrogel can scavenge ROS efficiently, enabling it to be a cell carrier of bone marrow-derived mesenchymal stem cells (BMSCs) to protect delivered cells from ROS-mediated death and osteogenesis inhibition, favorably enhancing the tissue repair potential of stem cells. Further in vivo investigations seriously demonstrated that this ROS-scavenging hydrogel encapsulated with BMSCs can prominently promote osteointegration of 3D printed microporous titanium alloy prosthesis in osteoporosis, including scavenging accumulated ROS, inducing macrophages to polarize toward M2 phenotype, suppressing inflammatory cytokines expression, and improving osteogenesis related markers (e.g., ALP, Runx-2, COL-1, BSP, OCN, and OPN). This work provides a novel strategy for conquering the challenge of transplanted stem cells cannot fully function in the impaired microenvironment, and enhancing prosthetic osteointegration in osteoporosis.

An ancestral role for 3-KETOACYL-COA SYNTHASE3 as a negative regulator of plant cuticular wax synthesis.

The plant cuticle, a structure primarily composed of wax and cutin, forms a continuous coating over most aerial plant surfaces. The cuticle plays important roles in plant tolerance to environmental stress, including stress imposed by drought. Some members of the 3-KETOACYL-COA SYNTHASE (KCS) family are known to act as metabolic enzymes involved in cuticular wax production. Here we report that Arabidopsis (Arabidopsis thaliana) KCS3, which was previously shown to lack canonical catalytic activity, instead functions as a negative regulator of wax metabolism by reducing the enzymatic activity of KCS6, a key KCS involved in wax production. We demonstrate that the role of KCS3 in regulating KCS6 activity involves physical interactions between specific subunits of the fatty acid elongation complex and is essential for maintaining wax homeostasis. We also show that the role of the KCS3-KCS6 module in regulating wax synthesis is highly conserved across diverse plant taxa from Arabidopsis to the moss Physcomitrium patens, pointing to a critical ancient and basal function of this module in finely regulating wax synthesis.

Intramedullary nail fixation assisted by locking plate for complex subtrochanteric femur fractures: A retrospective study.

BACKGROUND: This study aims to compare therapeutic effects of two methods in complicated subtrochanteric femur fractures surgery: intramedullary nail fixation assisted with lateral monocortical locking plate versus intramedullary nail fixation assisted with supplementary cables. METHODS: From June 2015 to June 2020, seventy-seven patients with complex subtrochanteric fractures (i.e., Seinsheimer's classification type IV or V) were included in this study. Thirty-six patients (plate group) were operated using the intramedullary nail fixation assisted by lateral monocortical locking plate, and forty-one patients (cable group) were using the intramedullary nail fixation assisted by cables. The clinical information and demographic results were collected and compared. RESULTS: Operation time of plate group was shorter than cable group and the Incisions length of plate group was longer. The fluoroscopy times were 22.8 ± 8.2 in plate group and 33.0 ± 9.0 in cable group (p < 0.01). Compared with the cable group, patients in plate group used less cerclage cables (p < 0.01). Patients in the plate group has less medial cortex displacement compared with the cable group. (p = 0.038). As for the angular difference of neck shaft angle between operated hip and uninjured hip, plate group has less difference compared with the cable group. Time to union was 14.2 ± 3.1 weeks in plate group which is shorter than the cable group (17.9 ± 4.8 weeks). In terms of follow up period, number of malunion, Harris hip score, walking ability and traumatic hip rating scale, no significant differences were detected. CONCLUSIONS: Our results suggest that using lateral monocortical plate as an auxiliary way may have a longer surgical incision and more intraoperative blood loss, however, the operation time is shorter, the fluoroscopy times is less, and the time to union is shorter. Intramedullary nail fixation assisted by lateral monocortical locking plate may be a new option for patients with complex subtrochanteric femur fractures.

Polydopamine and hyaluronic acid immobilisation on vancomycin-loaded titanium nanotube for prophylaxis of implant infections.

Titanium nanotube (Ti-NT) is an attractive substrate for local drug delivery, however, it is difficult to control the burst drug release and achieve sustained release from these nanotubes. In the present study, we investigated the feasibility of controlling drug release from Ti-NT within polydopamine and hyaluronic acid films, to achieve antibacterial activity and osteogenic promotion. Vancomycin was loaded into the Ti-NT by lyophilisation. Dopamine and hyaluronic acid were immobilized on the vancomycin-loaded Ti-NT surface through alternate deposition technique. The anti-infective and osteogenic abilities of the polydopamine and hyaluronic acid-modified Ti-NT were then investigated. Our results demonstrated that polydopamine and hyaluronic acid-modified Ti-NT exhibited improved drug loading and release control for 7 days. Compared with the vancomycin-loaded Ti-NT, the polydopamine and hyaluronic acid-modified Ti-NT exhibited better antibacterial ability, and the hyaluronic acid-modified Ti-NT promoted the osteogenic differentiation of rat bone marrow stem cells. Our results demonstrated that Ti-NT biofunctionalized with polydopamine and hyaluronic acid can help overcome the limitations of Ti-NT, by improving drug loading, antibacterial activity and osteogenic ability.

Deregulated expression and subcellular localization of CPSF6, a circRNA-binding protein, promote malignant development of esophageal squamous cell carcinoma.

Objective: Cleavage and polyadenylation specific factor 6 (CPSF6) has been documented as an oncoprotein in different types of cancer. However, functions of CPSF6 have not been investigated yet in esophageal squamous cell carcinoma (ESCC). Here, we aimed to investigate the potential clinical values and biological functions of CPSF6 in ESCC. Methods: For determining the expression level of CPSF6 in ESCC patients, we analyzed published data, performed quantitative real-time polymerase chain reaction (RT-qPCR) and immunohistochemistry assays. Kaplan-Meier curves and log-rank tests were used for survival analyses. GO and KEGG analyses were done for CPSF6-related genes. Cell proliferation, colony formation and xenograft assays were conducted to verify the effects of CPSF6 on ESCC. In addition, cell cycle and apoptosis assays were also performed to manifest the functions of CPSF6 and circCPSF6. RNA pulldown and radioimmunoprecipitation (RIP) assays were used for confirming the interaction between circCPSF6 (hsa_circ_0000417) and CPSF6 protein. The regulatory relationship between CPSF6 protein and circCPSF6 was determined by RT-qPCR. Results: We found that CPSF6 was upregulated in ESCC tissues and overexpression of cytoplasmic CPSF6 was associated with poor prognosis. GO and KEGG analyses suggested that CPSF6 could mainly affect cell division in ESCC. Further experiments manifested that CPSF6 promoted cell proliferation and colony formationin vitro. Xenograft assay showed that knockdown of CPSF6 significantly decreased tumor growth rate in vivo. Subsequently, we verified that depletion of CPSF6 led to cell cycle arrest and apoptosis. Finally, we validated that CPSF6, as a circRNA-binding protein, interacted with and regulated its circular isoform circCPSF6 (hsa_circ_0000417), of which depletion also resulted in cell cycle arrest and cell apoptosis in ESCC. Conclusions: These findings gave us insight that overexpression of cytoplasmic CPSF6 protein is associated with poor prognosis in ESCC and CPSF6 may function as an oncoprotein, at least in part, through regulating circCPSF6 expression.

Comparison of femoral nerve block and acupuncture analgesia for acute preoperative pain in elderly patients with femoral neck fracture: a retrospective study.

OBJECTIVE: This study was to compare the efficacy of femoral nerve block (FNB) and acupuncture for acute preoperative pain in patients with femoral neck fracture (FNF). METHODS: From June 2017 to June 2019, 130 patients with FNF were included in this study. Sixty-six patients received FNB treatment (FNB group) and sixty-four patients received acupuncture treatment (Acupuncture group). The clinical information, visual analog scale (VAS) scores, nursing quality scores, sleep quality scores, delirium numbers, and perioperative complications were collected and compared between the 2 groups. RESULTS: The resting VAS score and the exercise VAS score decreased after FNB or acupuncture in both groups. Thirty minutes after analgesia, the resting VAS scores in the FNB group and the acupuncture group were 27.3±8.0 and 27.9±7.8, respectively (P=0.67); while exercise VAS scores were 60.2±10.4 and 59.5±9.8, respectively (P=0.73). In addition, there was no statistical difference in the VAS score between the two groups on day 1 and day 2 after admission. There was no statistical difference in nursing quality, sleep rhythm disorder, sleep quality, or times of mental disorder between the two groups. CONCLUSION: FNB analgesia and acupuncture analgesia are safe and effective for the control of acute preoperative pain in senile patients with femoral neck fracture. Both methods have good analgesic effects, which can improve nursing and sleep quality, and reduce the incidence of delirium. As a traditional Chinese medicine method, acupuncture analgesia can effectively manage the acute preoperative pain in senile femoral neck fracture patients.

Cytology, transcriptomics, and mass spectrometry imaging reveal changes in late-maturation elm (Ulmus pumila) seeds.

During seed maturation, the seed deposits storage compounds (starches, oils, and proteins), synthesizes defense compounds, produces a seed coat, initiates embryo dormancy, and becomes desiccated. During the late-maturation stage, seed storage compound contents and compositions change dramatically. Although maturation has been extensively studied in model species and crops, it remains less well characterized in woody perennial plants. In this study, we conducted morphological and cytological observations, transcriptome profiling, and chemical constituent analysis of elm (Ulmus pumila L.) seeds during the late-maturation stage. Light and electron microscopy revealed that closely packed yet discrete lipid bodies frequently surrounded the densely stained protein bodies, and the protein bodies became irregular or even partially disintegrated at the end of seed development. RNA-seq detected substantial transcriptome changes during the late-maturation stage, and pathway enrichment analysis showed that the differentially expressed genes were associated with phenylpropanoid biosynthesis, starch and sucrose metabolism, plant-pathogen interactions, and hormone signal transduction. Furthermore, we used mass spectrometry imaging to detect the relative intensity and spatial distribution of fatty acids, phospholipids, and waxes in elm seeds. Our findings provide a framework for understanding the changes in cytological features and chemical composition during the final stage of elm seed development, and a detailed reference for seed development in woody plants.

Multi-instance learning of graph neural networks for aqueous pKa prediction.

MOTIVATION: The acid dissociation constant (pKa) is a critical parameter to reflect the ionization ability of chemical compounds and is widely applied in a variety of industries. However, the experimental determination of pKa is intricate and time-consuming, especially for the exact determination of micro-pKa information at the atomic level. Hence, a fast and accurate prediction of pKa values of chemical compounds is of broad interest. RESULTS: Here, we compiled a large-scale pKa dataset containing 16 595 compounds with 17 489 pKa values. Based on this dataset, a novel pKa prediction model, named Graph-pKa, was established using graph neural networks. Graph-pKa performed well on the prediction of macro-pKa values, with a mean absolute error around 0.55 and a coefficient of determination around 0.92 on the test dataset. Furthermore, combining multi-instance learning, Graph-pKa was also able to automatically deconvolute the predicted macro-pKa into discrete micro-pKa values. AVAILABILITY AND IMPLEMENTATION: The Graph-pKa model is now freely accessible via a web-based interface (https://pka.simm.ac.cn/). SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.