RESUMO
Graph neural networks (GNNs) have demonstrated exceptional performance in processing various types of graph data, such as citation networks and social networks, etc. Although many of these GNNs prove their superiority in handling homophilic graphs, they often overlook the other kind of widespread heterophilic graphs, in which adjacent nodes tend to have different classes or dissimilar features. Recent methods attempt to address heterophilic graphs from the graph spatial domain, which try to aggregate more similar nodes or prevent dissimilar nodes with negative weights. However, they may neglect valuable heterophilic information or extract heterophilic information ineffectively, which could cause poor performance of downstream tasks on heterophilic graphs, including node classification and graph classification, etc. Hence, a novel framework named GARN is proposed to effectively extract both homophilic and heterophilic information. First, we analyze the shortcomings of most GNNs in tackling heterophilic graphs from the perspective of graph spectral and spatial theory. Then, motivated by these analyses, a Graph Aggregating-Repelling Convolution (GARC) mechanism is designed with the objective of fusing both low-pass and high-pass graph filters. Technically, it learns positive attention weights as a low-pass filter to aggregate similar adjacent nodes, and learns negative attention weights as a high-pass filter to repel dissimilar adjacent nodes. A learnable integration weight is used to adaptively fuse these two filters and balance the proportion of the learned positive and negative weights, which could control our GARC to evolve into different types of graph filters and prevent it from over-relying on high intra-class similarity. Finally, a framework named GARN is established by simply stacking several layers of GARC to evaluate its graph representation learning ability on both the node classification and image-converted graph classification tasks. Extensive experiments conducted on multiple homophilic and heterophilic graphs and complex real-world image-converted graphs indicate the effectiveness of our proposed framework and mechanism over several representative GNN baselines.
Assuntos
Redes Neurais de Computação , Algoritmos , Gráficos por ComputadorRESUMO
The synthesis of a specific product via the Fischer-Tropsch synthesis remains challenging due to the uncontrollable coupling of CHx on active sites. Isoparaffins, essential high-quality petroleum additives for improving octane numbers, are primarily derived from petroleum or natural gas. With petroleum reserves dwindling and the associated low selectivity, the direct conversion of syngas to isoparaffins has emerged as a promising alternative. This study presents a tandem catalyst comprising CoxMn1-xO and zeolites for catalyzing the direct conversion of syngas to C4-C5 isoparaffins. The relay catalyst exhibited an impressive selectivity of 55.6% toward the desired products while maintaining a low CO2 selectivity of approximately 20%. Notably, the selectivity of isobutane reached 43.5%, exceeding predictions based on the Anderson-Schulz-Flory distribution. Syngas undergoes conversion into olefins on CoxMn1-xO nanocomposites, diffuses into microporous zeolites, and interacts with Brønsted acids to produce isoparaffins. The stability of the relay catalyst relied significantly on the pore characteristics and acidic density of the zeolites.
RESUMO
Graph Convolutional Networks (GCNs) with naive message passing mechanisms have limited performance due to the isotropic aggregation strategy. To remedy this drawback, some recent works focus on how to design anisotropic aggregation strategies with tricks on feature mapping or structure mining. However, these models still suffer from the low ability of expressiveness and long-range modeling for the needs of high performance in practice. To this end, this paper proposes a tree-guided anisotropic GCN, which applies an anisotropic aggregation strategy with competitive expressiveness and a large receptive field. Specifically, the anisotropic aggregation is decoupled into two stages. The first stage is to establish the path of the message passing on a tree-like hypergraph consisting of substructures. The second one is to aggregate the messages with constrained intensities by employing an effective gating mechanism. In addition, a novel anisotropic readout mechanism is constructed to generate representative and discriminative graph-level features for downstream tasks. Our model outperforms baseline methods and recent works on several synthetic benchmarks and datasets from different real-world tasks. In addition, extensive ablation studies and theoretical analyses indicate the effectiveness of our proposed method.
Assuntos
Redes Neurais de ComputaçãoRESUMO
Introduction: Endophytes are colonizers of healthy plants and they normally exhibit biocontrol activities, such as reducing the occurrence of plant diseases and promoting plant growth. The endophytic bacterium Bacillus halotolerans Q2H2 (Q2H2) was isolated from the roots of potato plants and was found to have an antagonistic effect on pathogenic fungi. Methods: Q2H2 was identified by morphological observations, physiological and biochemical identification, and 16S rRNA gene sequence analysis. Genes related to the anti-fungal and growth-promoting effects were analyzed using whole-genome sequencing and comparative genomic analysis. Finally, we analyzed the growth-promoting and biocontrol activities of Q2H2 in potato plants using pot experiments. Results: Antagonism and non-volatile substance plate tests showed that Q2H2 had strong antagonism against Fusarium oxysporum, Fusarium commune, Fusarium graminearum, Fusarium brachygibbosum, Rhizoctonia solani and Stemphylium solani. The plate test showed that Q2H2 had the ability to produce proteases, cellulases, ß-1,3-glucanase, dissolved organic phosphate, siderophores, indole-3-acetic acid (IAA), ammonia and fix nitrogen. The suitable growth ranges of Q2H2 under different forms of abiotic stress were pH 5-9, a temperature of 15-30°C, and a salt concentration of 1-5%. Though whole-genome sequencing, we obtained sequencing data of approximately 4.16 MB encompassed 4,102 coding sequences. We predicted 10 secondary metabolite gene clusters related to antagonism and growth promotion, including five known products surfactin, bacillaene, fengycin, bacilysin, bacillibactin, and subtilosin A. Average nucleotide identity and comparative genomic analyses revealed that Q2H2 was Bacillus halotolerans. Through gene function annotation, we analyzed genes related to antagonism and plant growth promotion in the Q2H2 genome. These included genes involved in phosphate metabolism (pstB, pstA, pstC, and pstS), nitrogen fixation (nifS, nifU, salA, and sufU), ammonia production (gudB, rocG, nasD, and nasE), siderophore production (fhuC, fhuG, fhuB, and fhuD), IAA production (trpABFCDE), biofilm formation (tasA, bslA, and bslB), and volatile compound production (alsD, ilvABCDEHKY, metH, and ispE), and genes encoding hydrolases (eglS, amyE, gmuD, ganB, sleL, and ydhD). The potato pot test showed that Q2H2 had an obvious growth-promoting effect on potato roots and better control of Fusarium wilt than carbendazim. Conclusion: These findings suggest that the strain-specific genes identified in bacterial endophytes may reveal important antagonistic and plant growth-promoting mechanisms.
RESUMO
Introduction: Endophytes are non-pathogenic inhabitants of healthy plant tissues and have been found to promote plant growth and health. The endophytic bacterial strain Q2H1 was isolated from the roots of the potato and was identified to exhibit growth-promoting effects in potato plants. Methods: Whole-genome sequencing was performed to reveal the mechanism underlying its growth-promoting effect. The obtained sequencing data of approximately 5.65 MB encompassed 5,533 coding sequences. Of note, nine secondary metabolite gene clusters, including siderophore gene clusters, closely associated with plant growth promotion (PGP) were predicted by antiSMASH software. Comparative genomic analysis revealed that Q2H1 belongs to the genus Peribacillus. By gene function annotation, those genes related to plant growth-promoting activities, including indole-3-acetic acid (IAA) synthesis in tryptophan metabolism, siderophore biosynthetic activity, phosphate solubilization, nitrogen fixation, and related genes, were summarized. IAA (14.4 µg/ml) was presumptively produced by Q2H1 using the Salkowski colorimetric method. A total of five genes, namely, phoU, pstB, pstA1, pstC, and pstS, were annotated for phosphate solubilization, which is associated with the ability of the Q2H1 strain to solubilize phosphate under in vitro conditions. Results: It is revealed that genes in the Q2H1 genome associated with nitrogen fixation belonged to three groups, namely, nitrogen fixation (nifU, sufU, salA, and nifS), nitrogen metabolism (nirA, nrtB, and nasA), and glutamate synthesis (glnA, gltB, gltD, and gudB), supported by evidence that Q2H1 grew on medium without nitrogen. We have also identified a siderophore gene cluster located on the chromosome of Q2H1, including seven genes (viz., rbsR, rhbf, rhbE, rhbD, rhbC, rhbA, ddc, and an unknown gene). In the in vitro assay, a prominent brown circle around the colony was produced on the chrome azurol S medium at 48 and 72 h post-inoculation, indicating that the siderophore gene cluster in Q2H1 harbored the ability to produce siderophores. Conclusion: In summary, these findings implied that identifying strain-specific genes for their metabolic pathways in bacterial endophytes may reveal a variety of significant functions of plant growth-promoting mechanisms.
RESUMO
BACKGROUND: Brain tumor segmentation plays a significant role in clinical treatment and surgical planning. Recently, several deep convolutional networks have been proposed for brain tumor segmentation and have achieved impressive performance. However, most state-of-the-art models use 3D convolution networks, which require high computational costs. This makes it difficult to apply these models to medical equipment in the future. Additionally, due to the large diversity of the brain tumor and uncertain boundaries between sub-regions, some models cannot well-segment multiple tumors in the brain at the same time. RESULTS: In this paper, we proposed a lightweight hierarchical convolution network, called LHC-Net. Our network uses a multi-scale strategy which the common 3D convolution is replaced by the hierarchical convolution with residual-like connections. It improves the ability of multi-scale feature extraction and greatly reduces parameters and computation resources. On the BraTS2020 dataset, LHC-Net achieves the Dice scores of 76.38%, 90.01% and 83.32% for ET, WT and TC, respectively, which is better than that of 3D U-Net with 73.50%, 89.42% and 81.92%. Especially on the multi-tumor set, our model shows significant performance improvement. In addition, LHC-Net has 1.65M parameters and 35.58G FLOPs, which is two times fewer parameters and three times less computation compared with 3D U-Net. CONCLUSION: Our proposed method achieves automatic segmentation of tumor sub-regions from four-modal brain MRI images. LHC-Net achieves competitive segmentation performance with fewer parameters and less computation than the state-of-the-art models. It means that our model can be applied under limited medical computing resources. By using the multi-scale strategy on channels, LHC-Net can well-segment multiple tumors in the patient's brain. It has great potential for application to other multi-scale segmentation tasks.
Assuntos
Neoplasias Encefálicas , Humanos , Neoplasias Encefálicas/diagnóstico por imagem , Encéfalo , Neuroimagem , Incerteza , Processamento de Imagem Assistida por ComputadorRESUMO
Spatial-temporal graph modeling has been widely studied in many fields, such as traffic forecasting and energy analysis, where data has time and space properties. Existing methods focus on capturing stable and dynamic spatial correlations by constructing physical and virtual graphs along with graph convolution and temporal modeling. However, existing methods tending to smooth node features may obscure the spatial-temporal patterns among nodes. Worse, the graph structure is not always available in some fields, while the manually constructed stable or dynamic graphs cannot necessarily reflect the true spatial correlations either. This paper proposes a Subgraph-Aware Graph Structure Revision network (SAGSR) to overcome these limitations. Architecturally, a subgraph-aware structure revision graph convolution module (SASR-GCM) is designed, which revises the learned stable graph to obtain a dynamic one to automatically infer the dynamics of spatial correlations. Each of these two graphs is separated into one homophilic subgraph and one heterophilic subgraph by a subgraph-aware graph convolution mechanism, which aggregates similar nodes in the homophilic subgraph with positive weights, while keeping nodes with dissimilar features in the heterophilic subgraph mutually away with negative aggregation weights to avoid pattern obfuscation. By combining a gated multi-scale temporal convolution module (GMS-TCM) for temporal modeling, SAGSR can efficiently capture the spatial-temporal correlations and extract complex spatial-temporal graph features. Extensive experiments, conducted on two specific tasks: traffic flow forecasting and energy consumption forecasting, indicate the effectiveness and superiority of our proposed approach over several competitive baselines.
RESUMO
PURPOSE: This research was designed to analyze the clinical features and prognostic factors for surgical treatment of esophageal squamous cell carcinoma (ESCC) in elderly patients aged 70 years and over. METHODS: The clinical data and follow-up data of 68 ESCC patients aged 70 years and over were collected. The characteristics of surgical treatment, perioperative complications, overall survival (OS), and the factors affecting survival were analyzed. RESULTS: The incidence rate of postoperative complications was 36% and the mortality rate was 0% during and 90 days after surgery. The 5-year OS was 45.0% and the 5-year disease-free survival (DFS) was 38.0%. Univariate analysis showed that gender, Charlson Comorbidity Index (CCI), pathological type, tumor differentiation, depth of invasion, postoperative complications, and lymph node metastasis were the factors associated with OS. Multivariate analysis showed that pathologic type, depth of invasion, and lymph node metastasis were the independent predictors of OS. The ideal long-term survival in elderly patients with ESCC was achieved with radical resection. CONCLUSION: The pathological type and pathological stage were the important independent risk factors of prognosis.