Article
Computer Science, Artificial Intelligence
Achref Benarab, Jianguo Sun, Fahad Rafique, Allaoua Refoufi
Summary: This paper proposes a novel approach to learn global ontology entities embeddings by exploiting the structure and relationships in ontologies, yielding embeddings capturing the semantics and similarities in the source ontology. Three different neural network models based on two architectures have been proposed and evaluated, showing competitive results outperforming state-of-the-art ontology and word embedding models.
IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING
(2023)
Article
Biochemical Research Methods
Peiliang Lou, YuXin Dong, Antonio Jimeno Yepes, Chen Li
Summary: This study introduces a new bio-entity representation learning model ERBK, which encodes axioms and definitions using knowledge graph embedding method and deep convolutional neural network respectively. Experimental results show that ERBK outperforms existing methods in predicting protein-protein interactions and gene-disease associations, and maintains promising performance under zero-shot circumstances.
Article
Computer Science, Artificial Intelligence
Elena Sanchez-Nielsen, Francisco Chavez-Gutierrez
Summary: The traditional concept and practice of lawmaking is being transformed by the open government paradigm, which involves a new digital model that links transparency, participation, and collaboration. However, information system-enabled solutions for addressing crucial challenges in this context are still in their infancy, with none available that provide functionalities for semantic search and retrieval of video clips on political debates, support open participation and collaboration, and automatically generate outcomes in the lawmaking lifecycle.
Article
Biochemical Research Methods
Kwangmin Kim, Doheon Lee
Summary: In this paper, a concept recognition method for multi-token biological entities using neural models combined with literature contexts is proposed. The key aspect of the method is utilizing contextual information from biological knowledge-bases for concept normalization, followed by named entity recognition. The model showed improved performances over conventional methods, particularly for multi-token concepts with higher variations.
BMC BIOINFORMATICS
(2021)
Article
Computer Science, Artificial Intelligence
Alcides Lopes, Joel Carbonera, Daniela Schmidt, Luan Garcia, Fabricio Rodrigues, Mara Abel
Summary: This study proposes an approach to classify domain entities into top-level ontology concepts without the need for external knowledge resources. By combining the term representing the domain entity and its informal definition into a single text sentence, a deep neural network with a language model as a layer is used for classification. Experimental results show that using transformer-based language models achieves promising results in classifying domain entities into 82 top-level ontology concepts, with a micro F1-score of 94%.
KNOWLEDGE-BASED SYSTEMS
(2023)
Article
Biochemical Research Methods
Yue Cao, Yang Shen
Summary: The study introduces a Transformer-based protein function annotation model TALE, which achieves significant breakthroughs in applicability and generalizability by using only sequence information. Compared to other methods, TALE outperforms when only sequence input is available and demonstrates superior generalizability to new species, low similarity proteins, and rare functions.
Article
Computer Science, Artificial Intelligence
Alcides Goncalves Lopes Junior, Joel Luis Carbonera, Daniela Schimidt, Mara Abel
Summary: This paper proposes a deep learning approach that automatically classifies domain entities into top-level concepts using their informal definitions and word embeddings of terms representing them. Experimental results show that the proposed method outperforms baseline approaches by 6% in terms of F-score and is less affected by polysemy.
EXPERT SYSTEMS WITH APPLICATIONS
(2022)
Article
Biochemical Research Methods
Hasan Balci, Metin Can Siper, Nasim Saleh, Ilkin Safarli, Ludovic Roy, Merve Kilicarslan, Rumeysa Ozaydin, Alexander Mazein, Charles Auffray, Ozgun Babur, Emek Demir, Ugur Dogrusoz
Summary: Newt is a web-based tool developed for viewing, constructing and analyzing biological maps in standard formats like SBGN, SBML and SIF. It meets the considerable need for sophisticated pathway viewers and editors using the latest visualization techniques and libraries.
Article
Biochemical Research Methods
Mei Zuo, Yang Zhang
Summary: Motivation: Information about bacteria biotopes (BB) is crucial for microbiological research and applications. The BB task at BioNLP-OST 2019 focuses on extracting microorganism locations and phenotypes from biomedical texts. Our span-based model, utilizing a pre-trained BERT model, achieves significantly better performance in entity and relation extraction tasks for BBs compared to previous methods, showing a reduction of 21.6% in slot error rate (SER). The model also shows effectiveness in recognizing nested entities and can be applied to other related tasks with state-of-the-art performance.
Article
Computer Science, Information Systems
Lingping Kong, Varun Ojha, Ruobin Gao, Ponnuthurai Nagaratnam Suganthan, Vaclav Snasel
Summary: This study proposes a Global Representation (GR) based attention mechanism to alleviate the heterophily and over-smoothing issues. The model integrates geometric information and uses GR to construct the Key, discovering the relation between nodes and the structural representation of the graph. Experimental tests validate the performance of the proposed method and provide insights for future improvements.
INFORMATION SCIENCES
(2023)
Article
Biochemical Research Methods
Xiaoshuai Zhang, Lixin Wang, Hucheng Liu, Xiaofeng Zhang, Bo Liu, Yadong Wang, Junyi Li
Summary: Protein is essential in living organisms and understanding its function is crucial for drug discovery and disease treatment. In this article, the authors propose the Prot2GO model, which integrates protein sequence and PPI network data to predict protein functions. The model achieves state-of-the-art performance on multiple metrics.
IEEE-ACM TRANSACTIONS ON COMPUTATIONAL BIOLOGY AND BIOINFORMATICS
(2023)
Article
Computer Science, Artificial Intelligence
Anil Ramakrishna, Rahul Gupta, Shrikanth Narayanan
Summary: This study introduces a generative model for multi-dimensional annotation fusion, aiming to estimate the ground truth more accurately. The model is applicable to both global and time series annotation fusion problems, jointly modeling different dimensions with model parameters estimated using the Expectation-Maximization algorithm. The performance of the model is evaluated on synthetic data, real emotion corpora, and an artificial task with human annotations.
IEEE TRANSACTIONS ON AFFECTIVE COMPUTING
(2022)
Article
Biotechnology & Applied Microbiology
Lei Deng, Shengli Ren, Jingpu Zhang
Summary: This paper proposes a new computational method called DNGRGO, which is based on global heterogeneous networks, to predict the functions of lncRNAs. DNGRGO calculates the similarities among proteins, miRNAs, and lncRNAs, and annotates the functions of lncRNAs based on their similar protein-coding genes labeled with gene ontology (GO). Experimental results show that DNGRGO is able to annotate lncRNAs by capturing the low-dimensional features of the heterogeneous network, and integrating miRNA data can improve its predictive performance.
Article
Computer Science, Artificial Intelligence
Weizhuo Li, Qiu Ji, Songmao Zhang, Xuefeng Fu, Guilin Qi
Summary: Discovering semantic relationships among heterogeneous ontologies has been a core research topic in the Semantic Web. This paper proposes a graph-based interactive mapping revision method to reduce manual efforts and improve efficiency. Experimental evaluation shows that this method is more time-saving compared to other methods in most cases.
EXPERT SYSTEMS WITH APPLICATIONS
(2023)
Article
Biochemical Research Methods
Narmada Sambaturu, Vaidehi Pusadkar, Sridhar Hannenhalli, Nagasuma Chandra
Summary: PathExt is a computational tool that can identify differentially active paths with a control, or most active paths without one, in an omics-integrated biological network. It can extract characteristic genes and pathways for studying systems effectively.
Article
Oncology
Basak Bahcivanci, Roshan Shafiha, Georgios V. Gkoutos, Animesh Acharjee
Summary: Liver cancer is the fourth leading cause of cancer-related death globally, with hepatocellular carcinoma (HCC) accounting for the majority of cases. However, current immunotherapy approaches are only partially effective due to the immunosuppressive nature of the tumor microenvironment (TME). This study aims to understand the TME in HCC and discover new immune markers for overcoming immunotherapy resistance.
Article
Radiology, Nuclear Medicine & Medical Imaging
Yu Wang, Yu Hong, Yue Wang, Xin Zhou, Xin Gao, Chenyan Yu, Jiaxi Lin, Lu Liu, Jingwen Gao, Minyue Yin, Guoting Xu, Xiaolin Liu, Jinzhou Zhu
Summary: This study evaluated the feasibility of automated multimodal machine learning in predicting esophageal variceal (EV) bleeding. By integrating endoscopic images and clinical variables, the study developed deep learning models and multimodal machine learning models, and compared them with existing clinical indices. The results showed that the multimodal machine learning models achieved higher accuracy and sensitivity, making them a useful tool for predicting EV bleeding.
JOURNAL OF DIGITAL IMAGING
(2023)
Review
Surgery
Vasileios Charalampakis, Victor Roth Cardoso, Alistair Sharples, Maha Khalid, Luke Dickerson, Tom Wiggins, Georgios Gkoutos, Olga Tucker, Paul Super, Martin Richardson, Rajwinder Nijjar, Rishi Singhal
Summary: Oesophageal perforation is a rare and serious condition, and early diagnosis and treatment are crucial for patient survival. In recent years, there has been a significant shift in the management of iatrogenic perforations, with a more liberal use of CT for early diagnosis and a higher rate of oesophageal stenting as the primary treatment option.
SURGICAL ENDOSCOPY AND OTHER INTERVENTIONAL TECHNIQUES
(2023)
Article
Biochemical Research Methods
Xiaolin Wang, Hongli Gao, Ren Qi, Ruiqing Zheng, Xin Gao, Bin Yu
Summary: This study proposes a novel clustering method called scBKAP, which addresses the issues of high dropout rate and curse of dimensionality in scRNA-seq data by utilizing an autoencoder network and a dimensionality reduction model MPDR. Comprehensive experiments on 21 public scRNA-seq datasets and simulated datasets demonstrate the superior performance of scBKAP over nine state-of-the-art single-cell clustering methods.
IEEE-ACM TRANSACTIONS ON COMPUTATIONAL BIOLOGY AND BIOINFORMATICS
(2023)
Article
Multidisciplinary Sciences
Haoyang Li, Juexiao Zhou, Zhongxiao Li, Siyuan Chen, Xingyu Liao, Bin Zhang, Ruochi Zhang, Yu Wang, Shiwei Sun, Xin Gao
Summary: Spatial transcriptomics technologies are utilized to analyze transcriptomes while preserving spatial information, providing high-resolution characterization of transcriptional patterns and tissue architecture reconstruction. Cellular heterogeneity plays a crucial role in deciphering spatial patterns of cell types, and various related methods have been proposed. In this study, we benchmarked 18 existing methods for cellular deconvolution using 50 real-world and simulated datasets, evaluating their accuracy, robustness, and usability. CARD, Cell2location, and Tangram showed the best performance for the cellular deconvolution task. Additionally, we provide decision-tree-style guidelines and recommendations for method selection and their additional features, facilitating users in choosing the optimal method for their specific needs. This comprehensive evaluation of 18 state-of-the-art methods for cellular deconvolution in spatial transcriptomics and the accompanying decision-tree-style guidelines and recommendations are valuable resources for researchers in this field.
NATURE COMMUNICATIONS
(2023)
Article
Multidisciplinary Sciences
Hind Alamro, Maha A. Thafar, Somayah Albaradei, Takashi Gojobori, Magbubah Essack, Xin Gao
Summary: Despite being the most common cause of dementia and impaired cognitive function, an effective treatment for Alzheimer's disease (AD) remains elusive. This study developed a computational method that combines multiple hub gene ranking methods, feature selection methods, and machine learning to identify biomarkers and targets for AD. The results showed that feature selection methods outperformed hub gene sets in prediction performance, and a small number of genes were able to accurately distinguish AD samples from healthy controls.
SCIENTIFIC REPORTS
(2023)
Article
Computer Science, Artificial Intelligence
Wanyun Zhou, Yufei Liu, Yingxin Li, Siqi Kong, Weilin Wang, Boyun Ding, Jiyun Han, Chaozhou Mou, Xin Gao, Juntao Liu
Summary: The paper introduces TriNet, a tri-fusion neural network for accurate prediction of anticancer peptides and antimicrobial peptides. TriNet utilizes three types of features and training modules to improve predictions. Experimental results demonstrate the superiority of TriNet compared to other methods.
Article
Biology
Chonghui Liu, Yan Zhang, Xin Gao, Guohua Wang
Summary: PACSI is an efficient method for identifying cell subpopulations associated with disease phenotypes.
Article
Biochemistry & Molecular Biology
Daniel S. Malawsky, Eva van Walree, Benjamin M. Jacobs, Teng Hiang Heng, Qin Qin Huang, Ataf H. Sabir, Saadia Rahman, Saghira Malik Sharif, Ahsan Khan, Masa Umicevic Mirkov, Hiroyuki Kuwahara, Xin Gao, Fowzan S. Alkuraya, Danielle Posthuma, William G. Newman, Christopher J. Griffiths, Rohini Mathur, David A. van Heel, Sarah Finer, Jared O'Connell, Hilary C. Martin
Summary: This study investigated the association between autozygosity and common diseases, and discovered an effective method to reduce confounding. The results suggest that autozygosity has significant impact on common diseases, especially for type 2 diabetes among British Pakistanis.
Article
Computer Science, Artificial Intelligence
Jiahao Gong, Sanyuan Zhao, Kin-Man Lam, Xin Gao, Jianbing Shen
Summary: Visible-infrared person re-identification (VI-ReID) is a challenging task for full-time intelligent surveillance systems due to the large cross-modal discrepancy. Existing methods suffer from heterogeneous structures and different spectra. To address this issue, we propose the Spectrum-Insensitive Data Augmentation (SIDA) strategy, which alleviates disturbance in visible and infrared spectra and forces the network to learn spectrum-irrelevant features. Our method achieves state-of-the-art performance on two visible-infrared cross-modal Re-ID datasets.
COMPUTER VISION AND IMAGE UNDERSTANDING
(2023)
Article
Multidisciplinary Sciences
Juexiao Zhou, Haoyang Li, Xingyu Liao, Bin Zhang, Wenjia He, Zhongxiao Li, Longxi Zhou, Xin Gao
Summary: Revoking personal private data is a basic human right, and the authors propose a solution called AFS to audit and revoke patients' private data from pre-trained deep learning models, enhancing privacy protection and data revocation rights in real-world intelligent healthcare.
NATURE COMMUNICATIONS
(2023)
Article
Multidisciplinary Sciences
Injeong Shim, Hiroyuki Kuwahara, NingNing Chen, Mais O. Hashem, Lama AlAbdi, Mohamed Abouelhoda, Hong-Hee Won, Pradeep Natarajan, Patrick T. Ellinor, Amit V. Khera, Xin Gao, Fowzan S. Alkuraya, Akl C. Fahed
Summary: Polygenic risk prediction can effectively predict the risk of cardiometabolic diseases in Arab population and is comparable to that in European-ancestry individuals. The polygenic scores are associated with disease severity and independent of conventional risk factors.
NATURE COMMUNICATIONS
(2023)
Review
Medicine, General & Internal
Jie Wang, Hongyu Chen, Zihuan Tang, Jinquan Zhang, Yuanwei Xu, Ke Wan, Kifah Hussain, Georgios Gkoutos, Yuchi Han, Yucheng Chen
Summary: This study systematically assessed the association of tafamidis treatment with outcomes in patients with transthyretin amyloid cardiomyopathy (ATTR-CM). The results showed that tafamidis treatment had a positive impact on the outcomes of patients with ATTR-CM, reducing the risk of adverse cardiovascular events and all-cause death.
Article
Multidisciplinary Sciences
Winnie Chua, Victor R. Cardoso, Eduard Guasch, Moritz F. Sinner, Christoph Al-Taie, Paul Brady, Barbara Casadei, Harry J. G. M. Crijns, Elton A. M. P. Dudink, Stephane N. Hatem, Stefan Kaeaeb, Peter Kastner, Lluis Mont, Frantisek Nehaj, Yanish Purmah, Jasmeet S. Reyat, Ulrich Schotten, Laura C. Sommerfeld, Stef Zeemering, Andre Ziegler, Georgios V. Gkoutos, Paulus Kirchhof, Larissa Fabritz
Summary: Early detection of atrial fibrillation through the measurement of circulating biomarkers can reduce the risk of stroke, cardiovascular death, and heart failure.
SCIENTIFIC REPORTS
(2023)
Article
Physics, Multidisciplinary
Dieter Maier, Thomas E. Exner, Anastasios G. Papadiamantis, Ammar Ammar, Andreas Tsoumanis, Philip Doganis, Ian Rouse, Luke T. Slater, Georgios V. Gkoutos, Nina Jeliazkova, Hilmar Ilgenfritz, Martin Ziegler, Beatrix Gerhard, Sebastian Kopetsky, Deven Joshi, Lee Walker, Claus Svendsen, Haralambos Sarimveis, Vladimir Lobaskin, Martin Himly, Jeaphianne van Rijn, Laurent Winckers, Javier Millan Acosta, Egon Willighagen, Georgia Melagraki, Antreas Afantitis, Iseult Lynch
Summary: This paper introduces the importance and objectives of the NanoCommons project, and summarizes its infrastructure - the NanoCommons Knowledge Base, describing its features and functions. By connecting nanosafety data sources and tools, this knowledge base provides users with a user-friendly interface and API to access state-of-the-art tools for nanomaterial safety prediction, design, and risk assessment. The article also presents the relationship between the knowledge base and other initiatives and projects, as well as its application in the FAIRification of experimental workflows.
FRONTIERS IN PHYSICS
(2023)