RESUMO
Biomedical relation extraction (BioRE) is the task of automatically extracting and classifying relations between two biomedical entities in biomedical literature. Recent advances in BioRE research have largely been powered by supervised learning and large language models (LLMs). However, training of LLMs for BioRE with supervised learning requires human-annotated data, and the annotation process often accompanies challenging and expensive work. As a result, the quantity and coverage of annotated data are limiting factors for BioRE systems. In this paper, we present our system for the BioCreative VII challenge-DrugProt track, a BioRE system that leverages a language model structure and weak supervision. Our system is trained on weakly labelled data and then fine-tuned using human-labelled data. To create the weakly labelled dataset, we combined two approaches. First, we trained a model on the original dataset to predict labels on external literature, which will become a model-labelled dataset. Then, we refined the model-labelled dataset using an external knowledge base. Based on our experiment, our approach using refined weak supervision showed significant performance gain over the model trained using standard human-labelled datasets. Our final model showed outstanding performance at the BioCreative VII challenge, achieving 3rd place (this paper focuses on our participating system in the BioCreative VII challenge). Database URL: http://wonjin.info/biore-yoon-et-al-2022.
Assuntos
Pesquisa Biomédica , Idioma , Humanos , Bases de Dados FactuaisRESUMO
Graph Neural Networks (GNNs) have been widely applied to various fields due to their powerful representations of graph-structured data. Despite the success of GNNs, most existing GNNs are designed to learn node representations on the fixed and homogeneous graphs. The limitations especially become problematic when learning representations on a misspecified graph or a heterogeneous graph that consists of various types of nodes and edges. To address these limitations, we propose Graph Transformer Networks (GTNs) that are capable of generating new graph structures, which preclude noisy connections and include useful connections (e.g., meta-paths) for tasks, while learning effective node representations on the new graphs in an end-to-end fashion. We further propose enhanced version of GTNs, Fast Graph Transformer Networks (FastGTNs), that improve scalability of graph transformations. Compared to GTNs, FastGTNs are up to 230× and 150× faster in inference and training, and use up to 100× and 148× less memory while allowing the identical graph transformations as GTNs. In addition, we extend graph transformations to the semantic proximity of nodes allowing non-local operations beyond meta-paths. Extensive experiments on both homogeneous graphs and heterogeneous graphs show that GTNs and FastGTNs with non-local operations achieve the state-of-the-art performance for node classification tasks. The code is available: https://github.com/seongjunyun/Graph_Transformer_Networks.
Assuntos
Aprendizagem , Redes Neurais de Computação , SemânticaRESUMO
BACKGROUND: The shell of Haliotis diversicolor, or shijueming (SJM), is a type of traditional Chinese medicine. The SJM has appeared in historical records as early as the third and fourth centuries. Historical records have revealed that SJM had mainly been used to treat eye diseases. After the Qing Dynasty (1757), records had emerged, detailing the use of SJM for treating skin injuries, particularly for treating poorly managed ulcers or traumatic wounds. Furthermore, in our anti-inflammation-screening system, SJM significantly inhibited the expression of pro-inflammatory proteins. Previous studies have yet to adopt an animal model to verify the phenomenon and described in the historical records regarding the efficacy of SJM in promoting wound healing. Besides, the mechanism of wound healing effect of SJM is also not clear. METHODS: This study applied in vitro and in vivo models, tissue section analysis, and western blotting to evaluate the effect of SJM on wound healing. The RAW 264.7 cells were used in anti-inflammatory activity assay and phagocytic assay. Male Wistar rats were used to evaluate the effect of SJM on burn injury healing. A copper block (2 × 2 cm, 150 g) preheated to 165 °C in a dry bath was used to contact the skin area for 10 s, thus creating a full-thickness burn injury. The results were analyzed by hematoxylin and eosin staining, picrosirius red staining and Western blotting. RESULTS: The results revealed that in the in vitro model, the presence of SJM decreased the inducible nitric oxide synthase (iNOS) expression and enhanced the functions of macrophages. The results of the rat burn injury model revealed that SJM decreased neutrophil infiltration, promoted wound healing, thus increasing the collagen I content and promoting the expression of transforming growth factor-beta 1 (TGF-ß1) protein. We speculate that the effect and mechanism of SJM on promoting wound healing is related to macrophage activation. In the inflammation phase, SJM alleviates inflammation by inhibiting iNOS expression and removing neutrophils through phagocytosis. Furthermore, SJM induces the secretion of TGF-ß1, converting collagen during the tissue remodeling phase. CONCLUSIONS: According to our review of relevant literature, this is the first study that applied an evidence-based method to verify that SJM alleviates inflammation, enhances phagocytosis, and triggers wound healing after burn injury. The study findings reveal that SJM provides a promising therapeutic option for treating burn injury.
Assuntos
Exoesqueleto/química , Queimaduras/tratamento farmacológico , Gastrópodes/química , Cicatrização/efeitos dos fármacos , Animais , Linhagem Celular , Masculino , Medicina Tradicional Chinesa , Óxido Nítrico Sintase Tipo II/metabolismo , Fagocitose/efeitos dos fármacos , Ratos , Ratos WistarRESUMO
In this study, we screened compounds with skin whitening properties and favorable safety profiles from a series of marine related natural products, which were isolated from Formosan soft coral Cladiella australis. Our results indicated that 4-(phenylsulfanyl)butan-2-one could successfully inhibit pigment generation processes in mushroom tyrosinase platform assay, probably through the suppression of tyrosinase activity to be a non-competitive inhibitor of tyrosinase. In cell-based viability examinations, it demonstrated low cytotoxicity on melanoma cells and other normal human cells. It exhibited stronger inhibitions of melanin production and tyrosinase activity than arbutin or 1-phenyl-2-thiourea (PTU). Also, we discovered that 4-(phenylsulfanyl)butan-2-one reduces the protein expressions of melanin synthesis-related proteins, including the microphthalmia-associated transcription factor (MITF), tyrosinase-related protein-1 (Trp-1), dopachrome tautomerase (DCT, Trp-2), and glycoprotein 100 (GP100). In an in vivo zebrafish model, it presented a remarkable suppression in melanogenesis after 48 h. In summary, our in vitro and in vivo biological assays showed that 4-(phenylsulfanyl)butan-2-one possesses anti-melanogenic properties that are significant in medical cosmetology.
Assuntos
Butanonas/farmacologia , Melaninas/biossíntese , Melanossomas/metabolismo , Sulfetos/farmacologia , Animais , Butanonas/toxicidade , Sobrevivência Celular/efeitos dos fármacos , Técnicas In Vitro , Melanoma Experimental , Camundongos , Monofenol Mono-Oxigenase/antagonistas & inibidores , Sulfetos/toxicidade , Peixe-ZebraRESUMO
Visual data comprise of multi-scale and inhomogeneous signals. In this paper, we exploit these characteristics and develop a compact data representation technique based on a hierarchical tensor-based transformation. In this technique, an original multi-dimensional dataset is transformed into a hierarchy of signals to expose its multi-scale structures. The signal at each level of the hierarchy is further divided into a number of smaller tensors to expose its spatially inhomogeneous structures. These smaller tensors are further transformed and pruned using a tensor approximation technique. Our hierarchical tensor approximation supports progressive transmission and partial decompression. Experimental results indicate that our technique can achieve higher compression ratios and quality than previous methods, including wavelet transforms, wavelet packet transforms, and single-level tensor approximation. We have successfully applied our technique to multiple tasks involving multi-dimensional visual data, including medical and scientific data visualization, data-driven rendering and texture synthesis.