Article
Biochemical Research Methods
Zilong Hou, Yuning Yang, Hui Li, Ka-Chun Wong, Xiangtao Li
Summary: A novel computational method, iDeepSubMito, was proposed for predicting the location of mitochondrial proteins to the submitochondrial compartments. The method utilized ProteinELMo for encoding and a convolutional neural network architecture based on bidirectional LSTM with self-attention mechanism, outperforming other computational methods.
BRIEFINGS IN BIOINFORMATICS
(2021)
Article
Robotics
David J. Yoon, Haowei Zhang, Mona Gridseth, Hugues Thomas, Timothy D. Barfoot
Summary: The research presents an unsupervised parameter learning method in a Gaussian variational inference setting that combines classic trajectory estimation and deep learning for 3D lidar odometry. The framework shows superior performance in learning from lidar data and performs comparably to state-of-the-art ICP-based methods on the KITTI odometry dataset, with additional results from the Oxford RobotCar dataset.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2021)
Article
Biochemistry & Molecular Biology
Shihu Jiao, Quan Zou
Summary: A new predictor called iPVP-DRLF was developed to specifically and effectively identify plant vacuole proteins. By using hybrid features and the light gradient boosting machine algorithm, iPVP-DRLF outperforms other predictors in terms of accuracy. Experimental results also indicate that deep representation learning features have an advantage in the identification of plant vacuole proteins.
COMPUTATIONAL AND STRUCTURAL BIOTECHNOLOGY JOURNAL
(2022)
Article
Automation & Control Systems
Hanjiang Hu, Hesheng Wang, Zhe Liu, Weidong Chen
Summary: This paper proposes a coarse-to-fine localization method based on image retrieval, which uses multi-domain image translation and gradient-weighted similarity activation mapping loss to extract domain-invariant features and improve localization accuracy. Experiments demonstrate the effectiveness and strong generalization ability of the proposed method in challenging environments.
IEEE-CAA JOURNAL OF AUTOMATICA SINICA
(2022)
Article
Computer Science, Artificial Intelligence
Huan Tian, Bo Liu, Tianqing Zhu, Wanlei Zhou, Philip S. Yu
Summary: With the widespread deployment of learning models in daily life, researchers have found that many of these models generate discriminatory predictions towards sensitive attributes such as gender or race. To address this, a common approach is to learn fair features without sensitive information by removing features. However, this reduces the accuracy of the models. In this research, we propose CIFair, a method that can learn fair features without feature removal operations or task-irrelevant learning objectives. Experimental results on the CelebA dataset demonstrate that CIFair achieves better fair prediction results while maintaining model accuracy performance.
KNOWLEDGE-BASED SYSTEMS
(2023)
Article
Biochemistry & Molecular Biology
Bohui Li, Maarten Altelaar, Bas van Breukelen
Summary: In this study, a deep learning framework was used to integrate large-scale protein abundance and interaction data, successfully predicting protein-protein interactions and identifying 5010 protein complexes using a two-stage clustering strategy. The predicted complexes were found to be mostly ubiquitously expressed in all cell types and tissues, providing a comprehensive map of protein-protein interactions and a novel perspective on protein complexes.
INTERNATIONAL JOURNAL OF MOLECULAR SCIENCES
(2023)
Article
Biochemistry & Molecular Biology
Alexander Derry, Russ B. B. Altman
Summary: The identification and characterization of structural sites contributing to protein function are crucial but challenging. Existing methods for function prediction are limited, necessitating the development of new computational methods. COLLAPSE is a framework for learning deep representations of protein sites, which shows state-of-the-art performance on various tasks and provides a platform for computational protein analysis.
Article
Computer Science, Artificial Intelligence
Chiranjibi Sitaula, Yong Xiang, Sunil Aryal, Xuequan Lu
Summary: This study introduces the use of hybrid features to represent scene images, which differs from previous methods focusing solely on foreground or background information. By combining information from foreground, background, and hybrid sources, more accurate representation of scene images is achieved.
EXPERT SYSTEMS WITH APPLICATIONS
(2021)
Article
Robotics
Riccardo Mereu, Gabriele Trivigno, Gabriele Berton, Carlo Masone, Barbara Caputo
Summary: This study proposes a taxonomy of architectures for learning sequential descriptors for VPR and analyzes the strengths and weaknesses of these architectural choices through benchmarking experiments. The study also explores the feasibility of using Transformers instead of CNN backbones and introduces a new sequence-level aggregator.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2022)
Article
Food Science & Technology
Liangzhen Jiang, Jici Jiang, Xiao Wang, Yin Zhang, Bowen Zheng, Shuqi Liu, Yiting Zhang, Changying Liu, Yan Wan, Dabing Xiang, Zhibin Lv
Summary: This study developed a peptide sequence-based umami peptide predictor, iUP-BERT, using a deep learning pretrained neural network feature extraction method. After optimization, the model showed improved performance compared to existing methods. The built iUP-BERT web server can aid in improving the palatability of dietary supplements.
Article
Computer Science, Artificial Intelligence
Ran Su, Xiaoying Liu, Qiangguo Jin, Xiaofeng Liu, Leyi Wei
Summary: A predictive model named DeepRA was developed to accurately predict the molecular subtype and patient overall survival of GBM using deep imaging features and machine learning technologies. Experiments showed that DeepRA outperformed traditional hand-crafted methods and regular convolutional neural networks.
KNOWLEDGE-BASED SYSTEMS
(2021)
Article
Environmental Sciences
Jianwei Zhao, Qiang Zhai, Pengbo Zhao, Rui Huang, Hong Cheng
Summary: This paper proposes a novel approach for cross-view geo-localization by using generative techniques combined with transformers. The method effectively captures the co-visual relationships between aerial and ground views, achieving high accuracy and setting new records compared to existing models.
Article
Computer Science, Artificial Intelligence
Wanyin Wu, Dapeng Tao, Hao Li, Zhao Yang, Jun Cheng
Summary: This study summarizes different types of features and metric learning approaches for person re-identification from a label attributes perspective. By combining advanced methods in data enhancement and feature extraction, comprehensive experiments were conducted on metric learning methods using two datasets, revealing the relationships between loss functions, deep feature space, and metric learning.
PATTERN RECOGNITION
(2021)
Article
Biochemistry & Molecular Biology
Jici Jiang, Xinxu Lin, Yueqi Jiang, Liangzhen Jiang, Zhibin Lv
Summary: This study presents the development of a machine learning prediction method called iBitter-DRLF, based on deep learning techniques, to accurately identify bitter peptides. By utilizing deep representation learning, this method can make accurate predictions solely based on peptide sequence data. This is of significant importance for improving the palatability of peptide therapeutics and dietary supplements.
INTERNATIONAL JOURNAL OF MOLECULAR SCIENCES
(2022)
Article
Robotics
Kohei Honda, Kenji Koide, Masashi Yokozuka, Shuji Oishi, Atsuhiko Banno
Summary: This letter introduces a LiDAR odometry estimation framework that improves position estimation accuracy by seamlessly fusing various local geometric shapes, and experimental results show that it reduces relative trajectory errors compared to other methods.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2022)
Review
Biotechnology & Applied Microbiology
Chunyan Ao, Liang Yu, Quan Zou
Summary: This review comprehensively summarizes the predictors for protein, RNA, and DNA modification sites and their association with diseases, emphasizing the importance of accurately identifying and understanding modification sites for disease research.
BRIEFINGS IN FUNCTIONAL GENOMICS
(2021)
Article
Biochemistry & Molecular Biology
Mengting Niu, Yuan Lin, Quan Zou
Summary: The study introduced an ensemble convolutional neural network model to accurately predict high on-target sgRNA activity in four crops, demonstrating the importance of machine learning methods in guiding crop gene editing and academic research.
PLANT MOLECULAR BIOLOGY
(2021)
Article
Biochemical Research Methods
Chao Wang, Ying Ju, Quan Zou, Chen Lin
Summary: A novel tool, DeepAc4C, was developed to identify ac4C using convolutional neural networks, achieving better and more balanced performance than existing predictors. By evaluating the impact of specific features on model predictions and their interaction effects, several interesting sequence motifs specific to ac4C were identified.
Article
Biochemical Research Methods
Yansu Wang, Lei Xu, Quan Zou, Chen Lin
Summary: The study introduces a new computational approach called prPred-DRLF, which accurately predicts plant R proteins using deep representation learning models. The results show that prPred-DRLF outperforms traditional methods in plant R protein prediction tasks.
Article
Computer Science, Information Systems
Qian Zhao, Jiaqi Ma, Yu Wang, Fang Xie, Zhibin Lv, Yaoqun Xu, Hua Shi, Ke Han
Summary: SNO is crucial for plant immune response and human disease treatment, with the efficient prediction tool Mul-SNO showing promising results.
IEEE JOURNAL OF BIOMEDICAL AND HEALTH INFORMATICS
(2022)
Article
Biochemistry & Molecular Biology
Jici Jiang, Xinxu Lin, Yueqi Jiang, Liangzhen Jiang, Zhibin Lv
Summary: This study presents the development of a machine learning prediction method called iBitter-DRLF, based on deep learning techniques, to accurately identify bitter peptides. By utilizing deep representation learning, this method can make accurate predictions solely based on peptide sequence data. This is of significant importance for improving the palatability of peptide therapeutics and dietary supplements.
INTERNATIONAL JOURNAL OF MOLECULAR SCIENCES
(2022)
Article
Genetics & Heredity
Mingxin Li, Yu Fan, Yiting Zhang, Zhibin Lv
Summary: The research focused on the impact of different feature information of miRNA sequences on the relationship between miRNA and disease. It found that a better graph neural network prediction model of miRNA-disease relationship could be built using CKSNAP feature, and the predicted miRNAs related to lung tumors, esophageal tumors, and kidney tumors were consistent with the wet experiment validation database.
Article
Genetics & Heredity
Liangzhen Jiang, Changying Liu, Yu Fan, Qi Wu, Xueling Ye, Qiang Li, Yan Wan, Yanxia Sun, Liang Zou, Dabing Xiang, Zhibin Lv
Summary: - This study assessed the transcriptional dynamics of filling stage Tartary buckwheat seeds and identified key genes related to seed development through RNA sequencing. Phytohormones ABA, AUX, ET, BR and CTK, along with related TFs, were found to substantially regulate seed development by targeting downstream expansin genes and structural starch biosynthetic genes. The transcriptome data could serve as a theoretical basis for improving the yield of Tartary buckwheat.
FRONTIERS IN GENETICS
(2022)
Editorial Material
Genetics & Heredity
Zhibin Lv, Mingxin Li, Yansu Wang, Quan Zou
FRONTIERS IN GENETICS
(2023)
Article
Food Science & Technology
Liangzhen Jiang, Jici Jiang, Xiao Wang, Yin Zhang, Bowen Zheng, Shuqi Liu, Yiting Zhang, Changying Liu, Yan Wan, Dabing Xiang, Zhibin Lv
Summary: This study developed a peptide sequence-based umami peptide predictor, iUP-BERT, using a deep learning pretrained neural network feature extraction method. After optimization, the model showed improved performance compared to existing methods. The built iUP-BERT web server can aid in improving the palatability of dietary supplements.
Article
Chemistry, Multidisciplinary
Hongdi Pei, Jiayu Li, Shuhan Ma, Jici Jiang, Mingxin Li, Quan Zou, Zhibin Lv
Summary: Thermophilic proteins have the potential to be used as biocatalysts in biotechnology. BertThermo, a model using BERT as an automatic feature extraction tool, achieved high accuracy in identifying thermophilic proteins. It outperformed previous predictive algorithms and demonstrated robustness in various datasets.+
APPLIED SCIENCES-BASEL
(2023)
Article
Food Science & Technology
Jici Jiang, Jiayu Li, Junxian Li, Hongdi Pei, Mingxin Li, Quan Zou, Zhibin Lv
Summary: A deep learning method called iUmami-DRLF was developed to identify umami peptides based solely on peptide sequence information. The results show that deep learning significantly improved the capability of models in identifying umami peptides. This method can be used to further enhance the umami flavor of food for a satisfying umami-flavored diet.
Article
Biochemistry & Molecular Biology
Yiting Deng, Shuhan Ma, Jiayu Li, Bowen Zheng, Zhibin Lv
Summary: Anticancer peptides (ACPs) are a promising new therapeutic approach in cancer treatment, as they can selectively target cancer cells. This study utilized machine learning algorithms to predict potential ACP sequences based on physicochemical features extracted from peptide sequences. By using feature selection methods, 19 key amino acid physicochemical properties were identified that can predict the likelihood of a peptide sequence functioning as an ACP. The study aims to enhance the efficiency of designing peptide sequences for cancer treatment.
INTERNATIONAL JOURNAL OF MOLECULAR SCIENCES
(2023)
Article
Chemistry, Multidisciplinary
Jiayu Li, Jici Jiang, Hongdi Pei, Zhibin Lv
Summary: A new IL-10-induced peptide recognition method called IL10-Stack was introduced in this research, which utilized unified deep representation learning and a stacking algorithm. Feature extraction from peptide sequences was done using two approaches, Amino Acid Index (AAindex) and sequence-based unified representation (UniRep). The IL10-Stack model, constructed using a 1900-dimensional UniRep feature vector, demonstrated excellent performance in IL-10-induced peptide recognition with an accuracy of 0.910 and MCC of 0.820. Compared to existing methods, IL-10Pred and ILeukin10Pred, the IL10-Stack approach showed improved accuracy by 12.1% and 2.4% respectively. The IL10-Stack method has the potential to identify IL-10-induced peptides, aiding in the development of immunosuppressive drugs.
APPLIED SCIENCES-BASEL
(2023)
Review
Multidisciplinary Sciences
Jiayu Li, Shuhan Ma, Hongdi Pei, Jici Jiang, Quan Zou, Zhibin Lv
Summary: This review focuses on the development of Tcprs for solid tumor therapy and prognostic prediction, and proposes strategies to enhance CAR-T cells through targeting different Tcprs, which may lead to the development of a new generation of cell therapies.