Accurate prediction of molecular properties is crucial for selecting compounds with ideal properties and reducing the costs and risks of trials.Traditional methods based on manually crafted features and graph-based me...Accurate prediction of molecular properties is crucial for selecting compounds with ideal properties and reducing the costs and risks of trials.Traditional methods based on manually crafted features and graph-based methods have shown promising results in molecular property prediction.However,traditional methods rely on expert knowledge and often fail to capture the complex structures and interactions within molecules.Similarly,graph-based methods typically overlook the chemical structure and function hidden in molecular motifs and struggle to effectively integrate global and local molecular information.To address these limitations,we propose a novel fingerprint-enhanced hierarchical graph neural network(FH-GNN)for molecular property prediction that simultaneously learns information from hierarchical molecular graphs and fingerprints.The FH-GNN captures diverse hierarchical chemical information by applying directed message-passing neural networks(D-MPNN)on a hierarchical molecular graph that integrates atomic-level,motif-level,and graph-level information along with their relationships.Addi-tionally,we used an adaptive attention mechanism to balance the importance of hierarchical graphs and fingerprint features,creating a comprehensive molecular embedding that integrated hierarchical mo-lecular structures with domain knowledge.Experiments on eight benchmark datasets from MoleculeNet showed that FH-GNN outperformed the baseline models in both classification and regression tasks for molecular property prediction,validating its capability to comprehensively capture molecular informa-tion.By integrating molecular structure and chemical knowledge,FH-GNN provides a powerful tool for the accurate prediction of molecular properties and aids in the discovery of potential drug candidates.展开更多
Due to its synergistic effects and reduced side effects,combination therapy has become an important strategy for treating complex diseases.In traditional Chinese medicine(TCM),the“monarch,minister,assistant,envoy”co...Due to its synergistic effects and reduced side effects,combination therapy has become an important strategy for treating complex diseases.In traditional Chinese medicine(TCM),the“monarch,minister,assistant,envoy”compatibilities theory provides a systematic framework for drug compatibility and has guided the formation of a large number of classic formulas.However,due to the complex compositions and diverse mechanisms of action of TCM,it is difficult to comprehensively reveal its potential synergistic patterns using traditional methods.Synergistic prediction based on molecular compatibility theory provides new ideas for identifying combinations of active compounds in TCM.Compared to resource-intensive traditional experimental methods,artificial intelligence possesses the ability to mine synergistic patterns from multi-omics and structural data,providing an efficient means for modeling and optimizing TCM combinations.This paper systematically reviews the application progress of AI in the synergistic prediction of TCM active compounds and explores the challenges and prospects of its application in modeling combination relationships,thereby contributing to the modernization of TCM theory and methodological innovation.展开更多
Colorectal cancer(CRC)ranks among the top causes of cancer-related fatalities globally.Recent progress in genomics,proteomics,and bioinformatics has greatly improved our comprehension of the molecular underpinnings of...Colorectal cancer(CRC)ranks among the top causes of cancer-related fatalities globally.Recent progress in genomics,proteomics,and bioinformatics has greatly improved our comprehension of the molecular underpinnings of CRC,paving the way for targeted therapies and immunotherapies.Nonetheless,obstacles such as tumor heterogeneity and drug resistance persist,hindering advancements in treatment efficacy.In this context,the integration of artificial intelligence(AI)and organoid technology presents promising new avenues.AI can analyze genetic and clinical data to forecast disease risk,prognosis,and treatment responses,thereby expediting drug development and tailoring treatment plans.Organoids replicate the genetic traits and biological behaviors of tumors,acting as platforms for drug testing and the formulation of personalized treatment approaches.Despite notable strides in CRC research and treatment-from genetic insights to therapeutic innovations-numerous challenges endure,including the intricate tumor microen-vironment,tumor heterogeneity,adverse effects of immunotherapies,issues related to AI data quality and privacy,and the need for standardization in organoid culture.Future initiatives should concentrate on clarifying the pathogenesis of CRC,refining AI algorithms and organoid models,and creating more effective therapeutic strategies to alleviate the global impact of CRC.展开更多
Pancreatic cancer, particularly pancreatic ductal adenocarcinoma(PDAC), is one of the most lethal malignancies,which is characterized by a complex tumor microenvironment(TME) that fosters immune evasion and treatment ...Pancreatic cancer, particularly pancreatic ductal adenocarcinoma(PDAC), is one of the most lethal malignancies,which is characterized by a complex tumor microenvironment(TME) that fosters immune evasion and treatment resistance. Recent genomic advancements have unveiled diverse molecular subtypes of PDAC, providing insights into targeted therapies and precision medicine. This review synthesizes the current understanding of PDAC's molecular characterization and immunosuppressive TME, as well as emerging therapeutic strategies, including innovative approaches targeting key molecular pathways such as kirsten rat sarcoma viral oncogene homolog(KRAS), epidermal growth factor receptor(EGFR), and immune checkpoints. Despite advances, challenges remain in overcoming treatment resistance and inherent heterogeneity of pancreatic cancer subtypes. We highlight the need for multidisciplinary collaboration to enhance early diagnosis and develop individualized therapeutic protocols, paving the way for improving the outcomes of this aggressive cancer. This integrated perspective underscores the urgency of transforming the innovative research into pancreatic cancer management.展开更多
Advancements in molecular approaches have been utilized to breed crops with a wide range of economically valuable traits to develop superior cultivars.This review provides a concise overview of modern breakthroughs in...Advancements in molecular approaches have been utilized to breed crops with a wide range of economically valuable traits to develop superior cultivars.This review provides a concise overview of modern breakthroughs in molecular plant production.Genotyping and high-throughput phenotyping methods for predictive plant breeding are briefly discussed.In this study,we explore contemporary molecular breeding techniques for producing desirable crop varieties.These techniques include cisgenesis,clustered regularly interspaced short palindromic repeat(CRISPR/Cas9)gene editing,haploid induction,and de novo domestication.We examine the speed breeding approach-a strategy for cultivating plants under controlled conditions.We further highlight the significance of modern breeding technologies in efficiently utilizing agricultural resources for crop production in urban areas.The deciphering of crop genomes has led to the development of extensive DNA markers,quantitative trait loci(QTLs),and pangenomes associated with various desirable crop traits.This shift to the genotypic selection of crops considerably expedites the plant breeding process.Based on the plant population used,the connection between genotypic and phenotypic data provides several genetic elements,including genes,markers,and alleles that can be used in genomic breeding and gene editing.The integration of speed breeding with genomic-assisted breeding and cutting-edge genome editing tools has made it feasible to rapidly manipulate and generate multiple crop cycles and accelerate the plant breeding process.Breakthroughs in molecular techniques have led to substantial improvements in modern breeding methods.展开更多
This study identified castor oil and phosphate ester as effective retarders through setting time,tensile,and flexural tests,and determined their optimal dosages.The mechanism by which phosphate ester affects the setti...This study identified castor oil and phosphate ester as effective retarders through setting time,tensile,and flexural tests,and determined their optimal dosages.The mechanism by which phosphate ester affects the setting time of polyurethane was further investigated using molecular dynamics simulations.Fourier transform infrared spectroscopy was also employed to systematically study the physical and chemical interactions between phosphate esters and polyurethane materials.The results demonstrate that a 1%concentration of phosphate ester provides the most effective retarding effect with minimal impact on the strength of polyurethane.When phosphate ester is added to the B component of the two-component polyurethane system,its interaction energy with component A decreases,as do the diffusion coefficient and aggregation degree of component B on the surface of component A.This reduction in interaction slows the setting time.Additionally,the addition of phosphate ester to polyurethane leads to the disappearance or weakening of functional groups,indicating competitive interactions within the phosphate ester components that inhibit the reaction rate.展开更多
Recently,diffusion models have emerged as a promising paradigm for molecular design and optimization.However,most diffusion-based molecular generative models focus on modeling 2D graphs or 3D geom-etries,with limited ...Recently,diffusion models have emerged as a promising paradigm for molecular design and optimization.However,most diffusion-based molecular generative models focus on modeling 2D graphs or 3D geom-etries,with limited research on molecular sequence diffusion models.The International Union of Pure and Applied Chemistry(IUPAC)names are more akin to chemical natural language than the simplified molecular input line entry system(SMILES)for organic compounds.In this work,we apply an IUPAC-guided conditional diffusion model to facilitate molecular editing from chemical natural language to chemical language(SMILES)and explore whether the pre-trained generative performance of diffusion models can be transferred to chemical natural language.We propose DiffIUPAC,a controllable molecular editing diffusion model that converts IUPAC names to SMILES strings.Evaluation results demonstrate that our model out-performs existing methods and successfully captures the semantic rules of both chemical languages.Chemical space and scaffold analysis show that the model can generate similar compounds with diverse scaffolds within the specified constraints.Additionally,to illustrate the model’s applicability in drug design,we conducted case studies in functional group editing,analogue design and linker design.展开更多
Guided by molecular networking,nine novel curvularin derivatives(1-9)and 16 known analogs(10-25)were isolated from the hydrothermal vent sediment fungus Penicillium sp.HL-50.Notably,compounds 5-7 represented a hybrid ...Guided by molecular networking,nine novel curvularin derivatives(1-9)and 16 known analogs(10-25)were isolated from the hydrothermal vent sediment fungus Penicillium sp.HL-50.Notably,compounds 5-7 represented a hybrid of curvularin and purine.The structures and absolute configurations of compounds 1-9 were elucidated via nuclear magnetic resonance(NMR)spectroscopy,X-ray diffraction,electronic circular dichroism(ECD)calculations,^(13)C NMR calculation,modified Mosher's method,and chemical derivatization.Investigation of anti-inflammatory activities revealed that compounds 7-9,11,12,14,15,and 18 exhibited significant suppressive effects against lipopolysaccharide(LPS)-induced nitric oxide(NO)production in murine macrophage RAW264.7 cells,with IC_(50)values ranging from 0.44 to 4.40μmol·L^(-1).Furthermore,these bioactive compounds were found to suppress the expression of inflammation-related proteins,including inducible NO synthase(i NOS),cyclooxygenase-2(COX-2),NLR family pyrin domain-containing protein 3(NLRP3),and nuclear factor kappa-B(NF-κB).Additional studies demonstrated that the novel compound 7 possessed potent antiinflammatory activity by inhibiting the transcription of inflammation-related genes,downregulating the expression of inflammation-related proteins,and inhibiting the release of inflammatory cytokines,indicating its potential application in the treatment of inflammatory diseases.展开更多
Aqueous zinc-ion batteries(AZIBs)have regained interest due to their inherent safety and costeffectiveness.However,the zinc anode is notorious for side reactions and dendrite growth,which plague the practical applicat...Aqueous zinc-ion batteries(AZIBs)have regained interest due to their inherent safety and costeffectiveness.However,the zinc anode is notorious for side reactions and dendrite growth,which plague the practical application of AZIBs.Adjusting the interfacial pH to reduce the by-products has been proven to be effective in protecting the zinc anode.Nevertheless,the dynamic regulation of the inherently unstable zinc interface during prolonged cycling remains a significant challenge.Herein,zwitterionic N-tris(hydroxymethyl)methylglycine(TMG)integrated with negative-COO^(-)and positive NH_(2)^(+)groups is proposed to stabilize the Zn anode and extend the lifespan as a self-regulating interfacial additive.The anionic portion serves as a trapping site to balance the interfacial pH and thus mitigate the unintended side reactions.Simultaneously,the NH_(2)^(+)cations are anchored on the zinc surface,forming a water-shielding,zincophilic molecular layer that guides three-dimensional diffusion and promotes uniform electro-deposition.Thus,an average plating efficiency of 99.74%over 3300 cycles at a current density of2 mA cm^(-2)is achieved.Notably,the TMG additive actualizes ultralong life in Zn‖Zn symmetrical cells(5500 h,exceeding 229 days,1 mA cm^(-2)/1 mA h cm^(-2)),and enables the Zn‖I_(2)cells to reach capacity retention rate of 89.4%after 1000 cycles at 1 A g^(-1).展开更多
The pre-wetting of aggregate surface is a means to improve the interface performance of SBS modified asphalt and aggregate.The effect of pre-wetting technology on the interaction between SBS modified asphalt and aggre...The pre-wetting of aggregate surface is a means to improve the interface performance of SBS modified asphalt and aggregate.The effect of pre-wetting technology on the interaction between SBS modified asphalt and aggregate was analyzed by molecular dynamics simulation.The diffusion coefficient and concentration distribution of SBS modified asphalt on aggregate surface are included.The simulation results show that the diffusion coefficient of the aggregate surface of SBS modified asphalt is increased by 47.6%and 70.5%respectively after 110#asphalt and 130#asphalt are pre-wetted.The concentration distribution of SBS modified asphalt on the aggregate surface after pre-wetting is more uniform.According to the results of interface energy calculation,the interface energy of SBS modified bitumen and aggregate can be increased by about 5%after pre-wetting.According to the results of molecular dynamics simulation,the pre-wetting technology can effectively improve the interface workability of SBS modified bitumen and aggregate,so as to improve the interface performance.展开更多
Traditional Chinese medicine formula(TCMF)represents a fundamental component of Chinese medical practice,incorporating medical knowledge and practices from both Han Chinese and various ethnic minorities,while providin...Traditional Chinese medicine formula(TCMF)represents a fundamental component of Chinese medical practice,incorporating medical knowledge and practices from both Han Chinese and various ethnic minorities,while providing comprehensive insights into health and disease.The foundation of TCMF lies in its holistic approach,manifested through herbal compatibility theory,which has emerged from extensive clinical experience and evolved into a highly refined knowledge system.Within this framework,Chinese herbal medicines exhibit intricated characteristics,including multi-component interactions,diverse target sites,and varied biological pathways.These complexities pose significant challenges for understanding their molecular mechanisms.Contemporary advances in artificial intelligence(AI)are reshaping research in traditional Chinese medicine(TCM),offering immense potential to transform our understanding of the molecular mechanisms underlying TCMFs.This review explores the application of AI in uncovering these mechanisms,highlighting its role in compound absorption,distribution,metabolism,and excretion(ADME)prediction,molecular target identification,compound and target synergy recognition,pharmacological mechanisms exploration,and herbal formula optimization.Furthermore,the review discusses the challenges and opportunities in AI-assisted research on TCMF molecular mechanisms,promoting the modernization and globalization of TCM.展开更多
Carbon capture is an important strategy and is implemented to achieve the goals of CO_(2)reduction and carbon neutrality.As a high energy-efficient technology,membrane-based separation plays a crucial role in CO_(2)ca...Carbon capture is an important strategy and is implemented to achieve the goals of CO_(2)reduction and carbon neutrality.As a high energy-efficient technology,membrane-based separation plays a crucial role in CO_(2)capture.It is urgently needed for membrane-based CO_(2)capture to develop the high-performance membrane materials with high permeability,selectivity,and stability.Herein,ultrapermeable carbon molecular sieve(CMS)membranes are fabricated by py roly zing a finely-engineered benzoxazole-containing copolyimide precursor for efficient CO_(2)capture.The microstructure of CMS membrane has been optimized by initially engineering the precursor-chemistry and subsequently tuning the pyrolysis process.Deep insights into the structure-property relationship of CMSs are provided in detail by a combination of experimental characterization and molecular simulations.We demonstrate that the intrinsically high free volume environment of the precursor,coupled with the steric hindrance of thermostable contorted fragments,promotes the formation of loosely packed and ultramicroporous carbon structures within the resultant CMS membrane,thereby enabling efficient CO_(2)discrimination via size sieving and affinity.The membrane achieves an ultrahigh CO_(2)permeability,good selectivity,and excellent stability.After one month of long-term operation,the CO_(2)permeability in the mixed gas is maintained at 11,800 Barrer,with a CO_(2)/N_(2)selectivity exceeding 60.This study provides insights into the relationship between precursor-chemistry and CMS performance,and our ultrapermeable CMS membrane,which is scalable using thin film manufacturing,holds great potential for industrial CO_(2)capture.展开更多
Layer-structured Ruddlesden–Popper(RP)perovskites(RPPs)with decent stability have captured the imagination of the photovoltaic research community and bring hope for boosting the development of perovskite solar cell(P...Layer-structured Ruddlesden–Popper(RP)perovskites(RPPs)with decent stability have captured the imagination of the photovoltaic research community and bring hope for boosting the development of perovskite solar cell(PSC)technology.However,two-dimensional(2D)or quasi-2D RP PSCs are encountered with some challenges of the large exciton binding energy,blocked charge transport and poor film quality,which restrict their photovoltaic performance.Fortunately,these issues can be readily resolved by rationally designing spacer cations of RPPs.This review mainly focuses on how to design the molecular structures of organic spacers and aims to endow RPPs with outstanding photovoltaic applications.We firstly elucidated the important roles of organic spacers in impacting crystallization kinetics,charge transporting ability and stability of RPPs.Then we brought three aspects to attention for designing organic spacers.Finally,we presented the specific molecular structure design strategies for organic spacers of RPPs aiming to improve photovoltaic performance of RP PSCs.These proposed strategies in this review will provide new avenues to develop novel organic spacers for RPPs and advance the development of RPP photovoltaic technology for future applications.展开更多
The global energy demand is increasing rapidly,and it is imperative to develop shale hydrocarbon re-sources vigorously.The prerequisite for enhancing the exploitation efficiency of shale reservoirs is the systematic e...The global energy demand is increasing rapidly,and it is imperative to develop shale hydrocarbon re-sources vigorously.The prerequisite for enhancing the exploitation efficiency of shale reservoirs is the systematic elucidation of the occurrence characteristics,flow behavior,and enhanced oil recovery(EOR)mechanisms of shale oil within commonly developed nanopores.Molecular dynamics(MD)technique can simulate the occurrence,flow,and extraction processes of shale oil at the nanoscale,and then quantitatively characterize various fluid properties,flow characteristics,and action mechanisms under different reservoir conditions by calculating and analyzing a series of MD parameters.However,the existing review on the application of MD simulation in shale oil reservoirs is not systematic enough and lacks a summary of technical challenges and solutions.Therefore,recent MD studies on shale oil res-ervoirs were summarized and analyzed.Firstly,the applicability of force fields and ensembles of MD in shale reservoirs with different reservoir conditions and fluid properties was discussed.Subsequently,the calculation methods and application examples of MD parameters characterizing various properties of fluids at the microscale were summarized.Then,the application of MD simulation in the study of shale oil occurrence characteristics,flow behavior,and EOR mechanisms was reviewed,along with the elucidation of corresponding micro-mechanisms.Moreover,influencing factors of pore structure,wall properties,reservoir conditions,fluid components,injection/production parameters,formation water,and inorganic salt ions were analyzed,and some new conclusions were obtained.Finally,the main challenges associated with the application of MD simulations to shale oil reservoirs were discussed,and reasonable prospects for future MD research directions were proposed.The purpose of this review is to provide theoretical basis and methodological support for applying MD simulation to study shale oil reservoirs.展开更多
The complex morphological,anatomical,physiological,and chemical mechanisms within the aging brain have been the hot topic of research for centuries.The aging process alters the brain structure that affects functions a...The complex morphological,anatomical,physiological,and chemical mechanisms within the aging brain have been the hot topic of research for centuries.The aging process alters the brain structure that affects functions and cognitions,but the worsening of such processes contributes to the pathogenesis of neurodegenerative disorders,such as Alzheimer's disease.Beyond these observable,mild morphological shifts,significant functional modifications in neurotransmission and neuronal activity critically influence the aging brain.Understanding these changes is important for maintaining cognitive health,especially given the increasing prevalence of age-related conditions that affect cognition.This review aims to explore the age-induced changes in brain plasticity and molecular processes,differentiating normal aging from the pathogenesis of Alzheimer's disease,thereby providing insights into predicting the risk of dementia,particularly Alzheimer's disease.展开更多
The physiology of the central and enteric nervous systems and gastric muscle contributes to the complexities encountered in the research and clinical management of gastroparesis. A wide range of prescription drugs tar...The physiology of the central and enteric nervous systems and gastric muscle contributes to the complexities encountered in the research and clinical management of gastroparesis. A wide range of prescription drugs target the underlying neurotransmitter imbalances and adjust nucleotide levels in appropriate tissues, but treatment is unsatisfactory, as our understanding of the condition is far from complete. In this study, computational software is used to focus on the adenine nucleotide, ATP, as a comparative template for the structures of drugs used in gastroparesis treatment. The results demonstrate that muscarinic, dopamine, serotonin (5-HT) and histamine receptor ligand classes relate structurally and differentially to the molecular structure of ATP. In these neurotransmitter classes, compounds do not target cell membrane receptor G-protein signal transduction in a manner that provides a single mechanism for improving gastroparesis symptoms. The exploration of alternative nucleotide-based deficiencies of KATP channels, Na+/K+ATPases and guanine nucleotide directed nitrergic mechanisms should enhance our experimental approach to understanding this condition.展开更多
Since the discovery of carbon dots(CDs)in 2004,the unique photoluminescence phenomenon of CDs has attracted widespread attention.However,the molecular weight of CDs has not been adequately quantified at present,due to...Since the discovery of carbon dots(CDs)in 2004,the unique photoluminescence phenomenon of CDs has attracted widespread attention.However,the molecular weight of CDs has not been adequately quantified at present,due to CDs are atomically imprecise and their molecular weight distribution is broad.In this paper,a series of Pluronic-modified CDs were prepared and the structure of the CDs was briefly analyzed.Subsequently,a molecular weight measurement method based on colligative properties was developed,and the correction coefficient in the algorithm was briefly analyzed.The calculated molecular weight was applied to the determination of surface adsorption capacity.This work provided a method for averaging the molecular weight of atomically imprecise particulate materials,which is expected to provide new opportunities in related fields.展开更多
In Burkina Faso, as in other African countries, infertility has become a social burden for the population and a public health problem. Male infertility accounts for 30% to 40% of all infertility cases. The diagnosis o...In Burkina Faso, as in other African countries, infertility has become a social burden for the population and a public health problem. Male infertility accounts for 30% to 40% of all infertility cases. The diagnosis of male infertility or hypofertility is often made by a simple laboratory analysis of sperm to explore sperm parameters. In most African countries, such as Burkina Faso, microbiological analysis in the context of sperm analysis is still not developed, and is carried out solely based on microscopy and traditional culture, which does not allow the growth of fragile and demanding bacteria. Our study investigated the microorganisms of sperm that may be involved in male infertility, using conventional bacteriology techniques and real-time PCR. However, it did not intend to perform a multivariate statistical association analysis to estimate the association of microorganisms with abnormal semen parameters. This prospective cross-sectional pilot study was carried out on patients who visited the bacteriology laboratory of Centre MURAZ, a research Institute in Burkina Faso, for male infertility diagnosis between 2 August and 31 August 2021. Bacteria were isolated and identified using standard bacteriology techniques. In parallel, common pathogenic microorganisms known to be associated with male infertility were targeted and detected in the sperm using a multiplex real-time PCR assay. A total of 38 sperm samples were analyzed by bacteriological culture and bacteria isolated were Staphylococcus aureus (S. aureus) 5.55%, Klebsiella pneumoniae (K. pneumoniae), Enterococcus faecalis (E. faecalis), Streptococcus agalactiae (S. agalactiae) and Staphylococcus hoemalyticus (S. hoemalyticus) respectively 2.70%. Real-time PCR targeted and detected Chlamydia trachomatis (C. trachomatis) at 7.89%, Ureaplasma urealyticum (U. urealyticum) at 21.05%, Ureaplasma parvum (U. parvum) at 18.42%, Mycoplasma hominis (M. hominis) at 15.79%, Mycoplasma genitalium (M. genitalium) at 10.53% and Trichomonas vaginalis (T. vaginalis) at 2.63%. Neisseria gonorrhoeae (N. gonorrhoeae) was targeted by the real-time PCR assay and was not detected (0%) in the tested semen samples. Our study highlights critical limitations of culture performance (low sensitivity), particularly in Burkina Faso, which has a total inability to detect microorganisms (fragile and demanding microorganisms) detected by PCR-based assays. There is therefore an urgent need to at least optimize culture, procedures and algorithms for detection of microorganisms associated with male infertility in clinical laboratories of Burkina Faso. The most effective solution is the routine implementation of molecular diagnostic methods.展开更多
Predicting molecular properties is essential for advancing for advancing drug discovery and design. Recently, Graph Neural Networks (GNNs) have gained prominence due to their ability to capture the complex structural ...Predicting molecular properties is essential for advancing for advancing drug discovery and design. Recently, Graph Neural Networks (GNNs) have gained prominence due to their ability to capture the complex structural and relational information inherent in molecular graphs. Despite their effectiveness, the “black-box” nature of GNNs remains a significant obstacle to their widespread adoption in chemistry, as it hinders interpretability and trust. In this context, several explanation methods based on factual reasoning have emerged. These methods aim to interpret the predictions made by GNNs by analyzing the key features contributing to the prediction. However, these approaches fail to answer critical questions: “How to ensure that the structure-property mapping learned by GNNs is consistent with established domain knowledge”. In this paper, we propose MMGCF, a novel counterfactual explanation framework designed specifically for the prediction of GNN-based molecular properties. MMGCF constructs a hierarchical tree structure on molecular motifs, enabling the systematic generation of counterfactuals through motif perturbations. This framework identifies causally significant motifs and elucidates their impact on model predictions, offering insights into the relationship between structural modifications and predicted properties. Our method demonstrates its effectiveness through comprehensive quantitative and qualitative evaluations of four real-world molecular datasets.展开更多
基金supported by Macao Science and Technology Development Fund,Macao SAR,China(Grant No.:0043/2023/AFJ)the National Natural Science Foundation of China(Grant No.:22173038)Macao Polytechnic University,Macao SAR,China(Grant No.:RP/FCA-01/2022).
文摘Accurate prediction of molecular properties is crucial for selecting compounds with ideal properties and reducing the costs and risks of trials.Traditional methods based on manually crafted features and graph-based methods have shown promising results in molecular property prediction.However,traditional methods rely on expert knowledge and often fail to capture the complex structures and interactions within molecules.Similarly,graph-based methods typically overlook the chemical structure and function hidden in molecular motifs and struggle to effectively integrate global and local molecular information.To address these limitations,we propose a novel fingerprint-enhanced hierarchical graph neural network(FH-GNN)for molecular property prediction that simultaneously learns information from hierarchical molecular graphs and fingerprints.The FH-GNN captures diverse hierarchical chemical information by applying directed message-passing neural networks(D-MPNN)on a hierarchical molecular graph that integrates atomic-level,motif-level,and graph-level information along with their relationships.Addi-tionally,we used an adaptive attention mechanism to balance the importance of hierarchical graphs and fingerprint features,creating a comprehensive molecular embedding that integrated hierarchical mo-lecular structures with domain knowledge.Experiments on eight benchmark datasets from MoleculeNet showed that FH-GNN outperformed the baseline models in both classification and regression tasks for molecular property prediction,validating its capability to comprehensively capture molecular informa-tion.By integrating molecular structure and chemical knowledge,FH-GNN provides a powerful tool for the accurate prediction of molecular properties and aids in the discovery of potential drug candidates.
基金supported by the National Key Research and Development Program of China(No.2024YFC3506900)Science and Technology Program of Tianjin(No.24ZXZSSS00460)Special Project for Technological Innovation in New Productive Forces of Modern Chinese Medicines(No.24ZXZKSY00010)。
文摘Due to its synergistic effects and reduced side effects,combination therapy has become an important strategy for treating complex diseases.In traditional Chinese medicine(TCM),the“monarch,minister,assistant,envoy”compatibilities theory provides a systematic framework for drug compatibility and has guided the formation of a large number of classic formulas.However,due to the complex compositions and diverse mechanisms of action of TCM,it is difficult to comprehensively reveal its potential synergistic patterns using traditional methods.Synergistic prediction based on molecular compatibility theory provides new ideas for identifying combinations of active compounds in TCM.Compared to resource-intensive traditional experimental methods,artificial intelligence possesses the ability to mine synergistic patterns from multi-omics and structural data,providing an efficient means for modeling and optimizing TCM combinations.This paper systematically reviews the application progress of AI in the synergistic prediction of TCM active compounds and explores the challenges and prospects of its application in modeling combination relationships,thereby contributing to the modernization of TCM theory and methodological innovation.
基金Supported by the National Human Genetic Resources Sharing Service Platform,No.PT-2024-0303Qingdao Medical and Health Research Guidance Project,No.2023-WJZD202.
文摘Colorectal cancer(CRC)ranks among the top causes of cancer-related fatalities globally.Recent progress in genomics,proteomics,and bioinformatics has greatly improved our comprehension of the molecular underpinnings of CRC,paving the way for targeted therapies and immunotherapies.Nonetheless,obstacles such as tumor heterogeneity and drug resistance persist,hindering advancements in treatment efficacy.In this context,the integration of artificial intelligence(AI)and organoid technology presents promising new avenues.AI can analyze genetic and clinical data to forecast disease risk,prognosis,and treatment responses,thereby expediting drug development and tailoring treatment plans.Organoids replicate the genetic traits and biological behaviors of tumors,acting as platforms for drug testing and the formulation of personalized treatment approaches.Despite notable strides in CRC research and treatment-from genetic insights to therapeutic innovations-numerous challenges endure,including the intricate tumor microen-vironment,tumor heterogeneity,adverse effects of immunotherapies,issues related to AI data quality and privacy,and the need for standardization in organoid culture.Future initiatives should concentrate on clarifying the pathogenesis of CRC,refining AI algorithms and organoid models,and creating more effective therapeutic strategies to alleviate the global impact of CRC.
文摘Pancreatic cancer, particularly pancreatic ductal adenocarcinoma(PDAC), is one of the most lethal malignancies,which is characterized by a complex tumor microenvironment(TME) that fosters immune evasion and treatment resistance. Recent genomic advancements have unveiled diverse molecular subtypes of PDAC, providing insights into targeted therapies and precision medicine. This review synthesizes the current understanding of PDAC's molecular characterization and immunosuppressive TME, as well as emerging therapeutic strategies, including innovative approaches targeting key molecular pathways such as kirsten rat sarcoma viral oncogene homolog(KRAS), epidermal growth factor receptor(EGFR), and immune checkpoints. Despite advances, challenges remain in overcoming treatment resistance and inherent heterogeneity of pancreatic cancer subtypes. We highlight the need for multidisciplinary collaboration to enhance early diagnosis and develop individualized therapeutic protocols, paving the way for improving the outcomes of this aggressive cancer. This integrated perspective underscores the urgency of transforming the innovative research into pancreatic cancer management.
基金funded by the United Arab Emirates UniversityResearch Officegrant number 12F041 to KM。
文摘Advancements in molecular approaches have been utilized to breed crops with a wide range of economically valuable traits to develop superior cultivars.This review provides a concise overview of modern breakthroughs in molecular plant production.Genotyping and high-throughput phenotyping methods for predictive plant breeding are briefly discussed.In this study,we explore contemporary molecular breeding techniques for producing desirable crop varieties.These techniques include cisgenesis,clustered regularly interspaced short palindromic repeat(CRISPR/Cas9)gene editing,haploid induction,and de novo domestication.We examine the speed breeding approach-a strategy for cultivating plants under controlled conditions.We further highlight the significance of modern breeding technologies in efficiently utilizing agricultural resources for crop production in urban areas.The deciphering of crop genomes has led to the development of extensive DNA markers,quantitative trait loci(QTLs),and pangenomes associated with various desirable crop traits.This shift to the genotypic selection of crops considerably expedites the plant breeding process.Based on the plant population used,the connection between genotypic and phenotypic data provides several genetic elements,including genes,markers,and alleles that can be used in genomic breeding and gene editing.The integration of speed breeding with genomic-assisted breeding and cutting-edge genome editing tools has made it feasible to rapidly manipulate and generate multiple crop cycles and accelerate the plant breeding process.Breakthroughs in molecular techniques have led to substantial improvements in modern breeding methods.
基金Funded by the National Natural Science Foundation of China(No.52370128)the Fundamental Research Funds for the Central Universities(No.2572022AW54)。
文摘This study identified castor oil and phosphate ester as effective retarders through setting time,tensile,and flexural tests,and determined their optimal dosages.The mechanism by which phosphate ester affects the setting time of polyurethane was further investigated using molecular dynamics simulations.Fourier transform infrared spectroscopy was also employed to systematically study the physical and chemical interactions between phosphate esters and polyurethane materials.The results demonstrate that a 1%concentration of phosphate ester provides the most effective retarding effect with minimal impact on the strength of polyurethane.When phosphate ester is added to the B component of the two-component polyurethane system,its interaction energy with component A decreases,as do the diffusion coefficient and aggregation degree of component B on the surface of component A.This reduction in interaction slows the setting time.Additionally,the addition of phosphate ester to polyurethane leads to the disappearance or weakening of functional groups,indicating competitive interactions within the phosphate ester components that inhibit the reaction rate.
基金supported by the Yonsei University graduate school Department of Integrative Biotechnology.
文摘Recently,diffusion models have emerged as a promising paradigm for molecular design and optimization.However,most diffusion-based molecular generative models focus on modeling 2D graphs or 3D geom-etries,with limited research on molecular sequence diffusion models.The International Union of Pure and Applied Chemistry(IUPAC)names are more akin to chemical natural language than the simplified molecular input line entry system(SMILES)for organic compounds.In this work,we apply an IUPAC-guided conditional diffusion model to facilitate molecular editing from chemical natural language to chemical language(SMILES)and explore whether the pre-trained generative performance of diffusion models can be transferred to chemical natural language.We propose DiffIUPAC,a controllable molecular editing diffusion model that converts IUPAC names to SMILES strings.Evaluation results demonstrate that our model out-performs existing methods and successfully captures the semantic rules of both chemical languages.Chemical space and scaffold analysis show that the model can generate similar compounds with diverse scaffolds within the specified constraints.Additionally,to illustrate the model’s applicability in drug design,we conducted case studies in functional group editing,analogue design and linker design.
基金funded by the National Key Research and Development Program of China(No.2022YFC2804101)the Guangdong Provincial Key R&D Program(No.2023B1111050011)+2 种基金the Guangdong Basic and Applied Basic Research Foundation(No.2023A1515010432)the Guangzhou Basic and Applied Basic Research Foundation(No.202201010305)the High-Level Talents Special Program of Zhejiang(No.2022R52036)。
文摘Guided by molecular networking,nine novel curvularin derivatives(1-9)and 16 known analogs(10-25)were isolated from the hydrothermal vent sediment fungus Penicillium sp.HL-50.Notably,compounds 5-7 represented a hybrid of curvularin and purine.The structures and absolute configurations of compounds 1-9 were elucidated via nuclear magnetic resonance(NMR)spectroscopy,X-ray diffraction,electronic circular dichroism(ECD)calculations,^(13)C NMR calculation,modified Mosher's method,and chemical derivatization.Investigation of anti-inflammatory activities revealed that compounds 7-9,11,12,14,15,and 18 exhibited significant suppressive effects against lipopolysaccharide(LPS)-induced nitric oxide(NO)production in murine macrophage RAW264.7 cells,with IC_(50)values ranging from 0.44 to 4.40μmol·L^(-1).Furthermore,these bioactive compounds were found to suppress the expression of inflammation-related proteins,including inducible NO synthase(i NOS),cyclooxygenase-2(COX-2),NLR family pyrin domain-containing protein 3(NLRP3),and nuclear factor kappa-B(NF-κB).Additional studies demonstrated that the novel compound 7 possessed potent antiinflammatory activity by inhibiting the transcription of inflammation-related genes,downregulating the expression of inflammation-related proteins,and inhibiting the release of inflammatory cytokines,indicating its potential application in the treatment of inflammatory diseases.
基金supported by the open research fund of Songshan Lake Materials Laboratory(2023SLABFN18)the Anhui Provincial Natural Science Foundation(2308085QB46)+2 种基金the Scientific Research Foundation of Education Department of Anhui Province of China(2022AH010025,2023AH051109)the Key Research and Development Program of Anhui Province of China(2022l07020011)The open research fund of the Anhui Key Lab of Metal Material and Processing(RZ2200002901)。
文摘Aqueous zinc-ion batteries(AZIBs)have regained interest due to their inherent safety and costeffectiveness.However,the zinc anode is notorious for side reactions and dendrite growth,which plague the practical application of AZIBs.Adjusting the interfacial pH to reduce the by-products has been proven to be effective in protecting the zinc anode.Nevertheless,the dynamic regulation of the inherently unstable zinc interface during prolonged cycling remains a significant challenge.Herein,zwitterionic N-tris(hydroxymethyl)methylglycine(TMG)integrated with negative-COO^(-)and positive NH_(2)^(+)groups is proposed to stabilize the Zn anode and extend the lifespan as a self-regulating interfacial additive.The anionic portion serves as a trapping site to balance the interfacial pH and thus mitigate the unintended side reactions.Simultaneously,the NH_(2)^(+)cations are anchored on the zinc surface,forming a water-shielding,zincophilic molecular layer that guides three-dimensional diffusion and promotes uniform electro-deposition.Thus,an average plating efficiency of 99.74%over 3300 cycles at a current density of2 mA cm^(-2)is achieved.Notably,the TMG additive actualizes ultralong life in Zn‖Zn symmetrical cells(5500 h,exceeding 229 days,1 mA cm^(-2)/1 mA h cm^(-2)),and enables the Zn‖I_(2)cells to reach capacity retention rate of 89.4%after 1000 cycles at 1 A g^(-1).
基金Funded by the Research Funds of China University of Mining and Technology(No.102523215)。
文摘The pre-wetting of aggregate surface is a means to improve the interface performance of SBS modified asphalt and aggregate.The effect of pre-wetting technology on the interaction between SBS modified asphalt and aggregate was analyzed by molecular dynamics simulation.The diffusion coefficient and concentration distribution of SBS modified asphalt on aggregate surface are included.The simulation results show that the diffusion coefficient of the aggregate surface of SBS modified asphalt is increased by 47.6%and 70.5%respectively after 110#asphalt and 130#asphalt are pre-wetted.The concentration distribution of SBS modified asphalt on the aggregate surface after pre-wetting is more uniform.According to the results of interface energy calculation,the interface energy of SBS modified bitumen and aggregate can be increased by about 5%after pre-wetting.According to the results of molecular dynamics simulation,the pre-wetting technology can effectively improve the interface workability of SBS modified bitumen and aggregate,so as to improve the interface performance.
基金supported by the National Key R&D Program of China(No.2022YFC3502005)the three-year Action Plan for Shanghai TCM Development and Inheritance Program[No.ZY(2021-2023)-0401]the National Natural Science Foundation of China(No.82104521)。
文摘Traditional Chinese medicine formula(TCMF)represents a fundamental component of Chinese medical practice,incorporating medical knowledge and practices from both Han Chinese and various ethnic minorities,while providing comprehensive insights into health and disease.The foundation of TCMF lies in its holistic approach,manifested through herbal compatibility theory,which has emerged from extensive clinical experience and evolved into a highly refined knowledge system.Within this framework,Chinese herbal medicines exhibit intricated characteristics,including multi-component interactions,diverse target sites,and varied biological pathways.These complexities pose significant challenges for understanding their molecular mechanisms.Contemporary advances in artificial intelligence(AI)are reshaping research in traditional Chinese medicine(TCM),offering immense potential to transform our understanding of the molecular mechanisms underlying TCMFs.This review explores the application of AI in uncovering these mechanisms,highlighting its role in compound absorption,distribution,metabolism,and excretion(ADME)prediction,molecular target identification,compound and target synergy recognition,pharmacological mechanisms exploration,and herbal formula optimization.Furthermore,the review discusses the challenges and opportunities in AI-assisted research on TCMF molecular mechanisms,promoting the modernization and globalization of TCM.
基金financial support from the National Key R&D Program of China(2021YFB3801200)the National Natural Science Foundation of China(22278051,22178044,22308043)CNPC Innovation Found(2022DQ02-0608)。
文摘Carbon capture is an important strategy and is implemented to achieve the goals of CO_(2)reduction and carbon neutrality.As a high energy-efficient technology,membrane-based separation plays a crucial role in CO_(2)capture.It is urgently needed for membrane-based CO_(2)capture to develop the high-performance membrane materials with high permeability,selectivity,and stability.Herein,ultrapermeable carbon molecular sieve(CMS)membranes are fabricated by py roly zing a finely-engineered benzoxazole-containing copolyimide precursor for efficient CO_(2)capture.The microstructure of CMS membrane has been optimized by initially engineering the precursor-chemistry and subsequently tuning the pyrolysis process.Deep insights into the structure-property relationship of CMSs are provided in detail by a combination of experimental characterization and molecular simulations.We demonstrate that the intrinsically high free volume environment of the precursor,coupled with the steric hindrance of thermostable contorted fragments,promotes the formation of loosely packed and ultramicroporous carbon structures within the resultant CMS membrane,thereby enabling efficient CO_(2)discrimination via size sieving and affinity.The membrane achieves an ultrahigh CO_(2)permeability,good selectivity,and excellent stability.After one month of long-term operation,the CO_(2)permeability in the mixed gas is maintained at 11,800 Barrer,with a CO_(2)/N_(2)selectivity exceeding 60.This study provides insights into the relationship between precursor-chemistry and CMS performance,and our ultrapermeable CMS membrane,which is scalable using thin film manufacturing,holds great potential for industrial CO_(2)capture.
基金funding from National Science Foundation of China(52202337 and 22178015)the Young Taishan Scholars Program of Shandong Province(tsqn202211082)+1 种基金Natural Science Foundation of Shandong Province(ZR2023MB051)Independent Innovation Research Project of China University of Petroleum(East China)(22CX06023A).
文摘Layer-structured Ruddlesden–Popper(RP)perovskites(RPPs)with decent stability have captured the imagination of the photovoltaic research community and bring hope for boosting the development of perovskite solar cell(PSC)technology.However,two-dimensional(2D)or quasi-2D RP PSCs are encountered with some challenges of the large exciton binding energy,blocked charge transport and poor film quality,which restrict their photovoltaic performance.Fortunately,these issues can be readily resolved by rationally designing spacer cations of RPPs.This review mainly focuses on how to design the molecular structures of organic spacers and aims to endow RPPs with outstanding photovoltaic applications.We firstly elucidated the important roles of organic spacers in impacting crystallization kinetics,charge transporting ability and stability of RPPs.Then we brought three aspects to attention for designing organic spacers.Finally,we presented the specific molecular structure design strategies for organic spacers of RPPs aiming to improve photovoltaic performance of RP PSCs.These proposed strategies in this review will provide new avenues to develop novel organic spacers for RPPs and advance the development of RPP photovoltaic technology for future applications.
基金supported by the National Natural Science Foundation of China(52304021,52104022,52204031)the Natural Science Foundation of Sichuan Province(2022NSFSC0205,2024NSFSC0201,2023NSFSC0947)the National Science and Technology Major Projects of China(2017ZX05049006-010).
文摘The global energy demand is increasing rapidly,and it is imperative to develop shale hydrocarbon re-sources vigorously.The prerequisite for enhancing the exploitation efficiency of shale reservoirs is the systematic elucidation of the occurrence characteristics,flow behavior,and enhanced oil recovery(EOR)mechanisms of shale oil within commonly developed nanopores.Molecular dynamics(MD)technique can simulate the occurrence,flow,and extraction processes of shale oil at the nanoscale,and then quantitatively characterize various fluid properties,flow characteristics,and action mechanisms under different reservoir conditions by calculating and analyzing a series of MD parameters.However,the existing review on the application of MD simulation in shale oil reservoirs is not systematic enough and lacks a summary of technical challenges and solutions.Therefore,recent MD studies on shale oil res-ervoirs were summarized and analyzed.Firstly,the applicability of force fields and ensembles of MD in shale reservoirs with different reservoir conditions and fluid properties was discussed.Subsequently,the calculation methods and application examples of MD parameters characterizing various properties of fluids at the microscale were summarized.Then,the application of MD simulation in the study of shale oil occurrence characteristics,flow behavior,and EOR mechanisms was reviewed,along with the elucidation of corresponding micro-mechanisms.Moreover,influencing factors of pore structure,wall properties,reservoir conditions,fluid components,injection/production parameters,formation water,and inorganic salt ions were analyzed,and some new conclusions were obtained.Finally,the main challenges associated with the application of MD simulations to shale oil reservoirs were discussed,and reasonable prospects for future MD research directions were proposed.The purpose of this review is to provide theoretical basis and methodological support for applying MD simulation to study shale oil reservoirs.
文摘The complex morphological,anatomical,physiological,and chemical mechanisms within the aging brain have been the hot topic of research for centuries.The aging process alters the brain structure that affects functions and cognitions,but the worsening of such processes contributes to the pathogenesis of neurodegenerative disorders,such as Alzheimer's disease.Beyond these observable,mild morphological shifts,significant functional modifications in neurotransmission and neuronal activity critically influence the aging brain.Understanding these changes is important for maintaining cognitive health,especially given the increasing prevalence of age-related conditions that affect cognition.This review aims to explore the age-induced changes in brain plasticity and molecular processes,differentiating normal aging from the pathogenesis of Alzheimer's disease,thereby providing insights into predicting the risk of dementia,particularly Alzheimer's disease.
文摘The physiology of the central and enteric nervous systems and gastric muscle contributes to the complexities encountered in the research and clinical management of gastroparesis. A wide range of prescription drugs target the underlying neurotransmitter imbalances and adjust nucleotide levels in appropriate tissues, but treatment is unsatisfactory, as our understanding of the condition is far from complete. In this study, computational software is used to focus on the adenine nucleotide, ATP, as a comparative template for the structures of drugs used in gastroparesis treatment. The results demonstrate that muscarinic, dopamine, serotonin (5-HT) and histamine receptor ligand classes relate structurally and differentially to the molecular structure of ATP. In these neurotransmitter classes, compounds do not target cell membrane receptor G-protein signal transduction in a manner that provides a single mechanism for improving gastroparesis symptoms. The exploration of alternative nucleotide-based deficiencies of KATP channels, Na+/K+ATPases and guanine nucleotide directed nitrergic mechanisms should enhance our experimental approach to understanding this condition.
文摘Since the discovery of carbon dots(CDs)in 2004,the unique photoluminescence phenomenon of CDs has attracted widespread attention.However,the molecular weight of CDs has not been adequately quantified at present,due to CDs are atomically imprecise and their molecular weight distribution is broad.In this paper,a series of Pluronic-modified CDs were prepared and the structure of the CDs was briefly analyzed.Subsequently,a molecular weight measurement method based on colligative properties was developed,and the correction coefficient in the algorithm was briefly analyzed.The calculated molecular weight was applied to the determination of surface adsorption capacity.This work provided a method for averaging the molecular weight of atomically imprecise particulate materials,which is expected to provide new opportunities in related fields.
文摘In Burkina Faso, as in other African countries, infertility has become a social burden for the population and a public health problem. Male infertility accounts for 30% to 40% of all infertility cases. The diagnosis of male infertility or hypofertility is often made by a simple laboratory analysis of sperm to explore sperm parameters. In most African countries, such as Burkina Faso, microbiological analysis in the context of sperm analysis is still not developed, and is carried out solely based on microscopy and traditional culture, which does not allow the growth of fragile and demanding bacteria. Our study investigated the microorganisms of sperm that may be involved in male infertility, using conventional bacteriology techniques and real-time PCR. However, it did not intend to perform a multivariate statistical association analysis to estimate the association of microorganisms with abnormal semen parameters. This prospective cross-sectional pilot study was carried out on patients who visited the bacteriology laboratory of Centre MURAZ, a research Institute in Burkina Faso, for male infertility diagnosis between 2 August and 31 August 2021. Bacteria were isolated and identified using standard bacteriology techniques. In parallel, common pathogenic microorganisms known to be associated with male infertility were targeted and detected in the sperm using a multiplex real-time PCR assay. A total of 38 sperm samples were analyzed by bacteriological culture and bacteria isolated were Staphylococcus aureus (S. aureus) 5.55%, Klebsiella pneumoniae (K. pneumoniae), Enterococcus faecalis (E. faecalis), Streptococcus agalactiae (S. agalactiae) and Staphylococcus hoemalyticus (S. hoemalyticus) respectively 2.70%. Real-time PCR targeted and detected Chlamydia trachomatis (C. trachomatis) at 7.89%, Ureaplasma urealyticum (U. urealyticum) at 21.05%, Ureaplasma parvum (U. parvum) at 18.42%, Mycoplasma hominis (M. hominis) at 15.79%, Mycoplasma genitalium (M. genitalium) at 10.53% and Trichomonas vaginalis (T. vaginalis) at 2.63%. Neisseria gonorrhoeae (N. gonorrhoeae) was targeted by the real-time PCR assay and was not detected (0%) in the tested semen samples. Our study highlights critical limitations of culture performance (low sensitivity), particularly in Burkina Faso, which has a total inability to detect microorganisms (fragile and demanding microorganisms) detected by PCR-based assays. There is therefore an urgent need to at least optimize culture, procedures and algorithms for detection of microorganisms associated with male infertility in clinical laboratories of Burkina Faso. The most effective solution is the routine implementation of molecular diagnostic methods.
文摘Predicting molecular properties is essential for advancing for advancing drug discovery and design. Recently, Graph Neural Networks (GNNs) have gained prominence due to their ability to capture the complex structural and relational information inherent in molecular graphs. Despite their effectiveness, the “black-box” nature of GNNs remains a significant obstacle to their widespread adoption in chemistry, as it hinders interpretability and trust. In this context, several explanation methods based on factual reasoning have emerged. These methods aim to interpret the predictions made by GNNs by analyzing the key features contributing to the prediction. However, these approaches fail to answer critical questions: “How to ensure that the structure-property mapping learned by GNNs is consistent with established domain knowledge”. In this paper, we propose MMGCF, a novel counterfactual explanation framework designed specifically for the prediction of GNN-based molecular properties. MMGCF constructs a hierarchical tree structure on molecular motifs, enabling the systematic generation of counterfactuals through motif perturbations. This framework identifies causally significant motifs and elucidates their impact on model predictions, offering insights into the relationship between structural modifications and predicted properties. Our method demonstrates its effectiveness through comprehensive quantitative and qualitative evaluations of four real-world molecular datasets.
