Article
Biochemical Research Methods
Qiaoming Liu, Ximei Luo, Jie Li, Guohua Wang
Summary: In this study, the authors propose an evolutionary sparse imputation (ESI) algorithm for single-cell transcriptomes to address the dropout problem and reduce data noise in gene expression profiles. The ESI algorithm constructs a sparse representation model based on gene regulation relationships between cells and uses an optimization framework based on nondominated sorting genetics to iteratively search for the global optimal solution. The results show that scESI outperforms benchmark methods in simulated datasets and real scRNA-seq datasets, improving cell classification, trajectory reconstruction, and identification of differentially expressed genes.
BRIEFINGS IN BIOINFORMATICS
(2022)
Article
Chemistry, Multidisciplinary
Yi Fan, Yunhe Wang, Fuzhou Wang, Lei Huang, Yuning Yang, Ka-c. Wong, Xiangtao Li
Summary: Unsupervised clustering is crucial for identifying cell types from scRNA-seq data. The proposed DEPF framework addresses the issue of inconsistent clustering results by using a silhouette coefficient-based indicator and a bi-objective fruit fly optimization algorithm. Multiple experiments on real datasets validate the effectiveness of DEPF in identifying and interpreting single-cell molecular heterogeneity.
Review
Genetics & Heredity
Amos Tanay, Arnau Sebe-Pedros
Summary: A fundamental characteristic of animal multicellularity is the coexistence of functionally specialized cell types encoded by a single genome, regulated by mechanisms controlling access to genomic information. Single-cell genomics is emerging as a powerful tool to catalog cell types and gene regulatory programs in non-traditional model species. Phylogenetic integration of cell atlases can lead to the development of cell type evolution models and a phylogenetic taxonomy of cells.
TRENDS IN GENETICS
(2021)
Article
Multidisciplinary Sciences
Richard C. Tyser, Elmir Mahammadov, Shota Nakanoh, Ludovic Vallier, Antonio Scialdone, Shankar Srinivas
Summary: The single-cell transcriptional profile of a human embryo between 16 and 19 days after fertilization shows similarities and differences in gastrulation compared to mouse and non-human primate models. This study provides new insights into human development and offers valuable information for directed differentiation of human cells in vitro.
Article
Multidisciplinary Sciences
Bang Tran, Duc Tran, Hung Nguyen, Seungil Ro, Tin Nguyen
Summary: Unsupervised clustering of scRNA-seq data is crucial for identifying cell types, but the challenges posed by large numbers of cells, high-dimensional data, and high dropout rates are significant. We introduce a new method called scCAN that accurately segregates different cell types in large and sparse scRNA-seq data, outperforming other state-of-the-art methods in terms of accuracy and scalability.
SCIENTIFIC REPORTS
(2022)
Article
Biochemistry & Molecular Biology
Xin Shao, Haihong Yang, Xiang Zhuang, Jie Liao, Penghui Yang, Junyun Cheng, Xiaoyan Lu, Huajun Chen, Xiaohui Fan
Summary: scDeepSort is a pre-trained tool for cell-type annotation in single-cell transcriptomics using deep learning and a weighted graph neural network. It demonstrates high performance and robustness across multiple datasets, achieving an accuracy of 83.79%.
NUCLEIC ACIDS RESEARCH
(2021)
Article
Multidisciplinary Sciences
Kai Battenberg, S. Thomas Kelly, Radu Abu Ras, Nicola A. Hetherington, Makoto Hayashi, Aki Minoda
Summary: Single-cell RNA-sequencing analysis has gained popularity, and UniverSC is a universal tool for processing single-cell RNA-seq data on any platform. It provides a command-line tool, docker image, and containerized graphical application for consistent and comprehensive integration, comparison, and evaluation of data from various platforms. Additionally, a cross-platform application with a graphical user interface is available to address the bottleneck of data processing for researchers without bioinformatics expertise.
NATURE COMMUNICATIONS
(2022)
Article
Biochemical Research Methods
Malte D. Luecken, M. Buettner, K. Chaichoompu, A. Danese, M. Interlandi, M. F. Mueller, D. C. Strobl, L. Zappia, M. Dugas, M. Colome-Tatche, Fabian J. Theis
Summary: This study benchmarked 68 method and preprocessing combinations on 85 batches of gene expression data, highlighting the importance of highly variable gene selection in improving method performance. When dealing with complex integration tasks, scANVI, Scanorama, scVI, and scGen consistently performed well, while the performance of single-cell ATAC-sequencing integration was strongly influenced by the choice of feature space.
Article
Multidisciplinary Sciences
Kiya W. Govek, Patrick Nicodemus, Yuxuan Lin, Jake Crawford, Artur B. Saturnino, Hannah Cui, Kristi Zoga, Michael P. Hart, Pablo G. Camara
Summary: The authors present a computational approach for cell morphometry and multi-modal analysis, which is based on concepts from metric geometry. They demonstrate the utility of this approach in integrating cell morphology data into single-cell omics analyses. The approach involves building cell morphology latent spaces using metric geometry, which facilitate the integration of single-cell morphological data and inference of relations with other data.
NATURE COMMUNICATIONS
(2023)
Article
Biochemistry & Molecular Biology
Jacob C. Kimmel, David R. Kelley
Summary: scNym is a semi-supervised adversarial neural network that can transfer cell identity annotations between different experiments by learning rich representations of cell identities from both labeled and unlabeled datasets. It shows superior performance in transferring annotations across experiments and can synthesize information from multiple datasets to improve accuracy. Additionally, scNym models are well calibrated, interpretable, and can be enhanced with saliency methods.
Article
Biochemical Research Methods
Johannes Smolander, Sini Junttila, Mikko S. Venalainen, Laura L. Elo
Summary: The scShaper is a new trajectory inference method that generates a continuous smooth pseudo-time using an ensemble approach. It accurately infers various trigonometric trajectories and outperforms other methods in terms of accuracy of cell sorting and differentially expressed genes. The scShaper is a fast method with a few hyperparameters, making it a promising alternative to the principal curves method for linear pseudotemporal ordering.
Article
Biochemistry & Molecular Biology
Mengwei Li, Xiaomeng Zhang, Kok Siong Ang, Jingjing Ling, Raman Sethi, Nicole Yee Shin Lee, Florent Ginhoux, Jinmiao Chen
Summary: DISCO is an integrated database of single-cell omics data, offering an integrated cell atlas and harmonized metadata that users can utilize for comprehensive single-cell data analysis and exploration.
NUCLEIC ACIDS RESEARCH
(2022)
Article
Biochemical Research Methods
Dakota Y. Hawkins, Daniel T. Zuch, James Huth, Nahomie Rodriguez-Sastre, Kelley R. McCutcheon, Abigail Glick, Alexandra T. Lion, Christopher F. Thomas, Abigail E. Descoteaux, William Evan Johnson, Cynthia A. Bradham
Summary: In this study, a novel unsupervised algorithm called ICAT is proposed, which can accurately identify and resolve cell states in single-cell RNA sequencing experiments, especially in perturbation experiments, where cell states need to be matched and population substructure needs to be removed. Through validation using simulated and real datasets, it is shown that ICAT outperforms current integration workflows and is robust to various conditions. Empirical validation in a developmental model demonstrates that only ICAT can identify perturbation-unique cellular responses.
Article
Biochemical Research Methods
Qianqian Zhang, Xing Xu, Li Lin, Jian Yang, Xing Na, Xin Chen, Lingling Wu, Jia Song, Chaoyong Yang
Summary: Cilo-seq is a high-performance platform for single-cell RNA sequencing library construction. It utilizes digital microfluidics in a single device, enabling convenient single-cell isolation, efficient nucleic acid amplification, low-loss nucleic acid purification, and high-quality library preparation.
Article
Multidisciplinary Sciences
Lieke Michielsen, Marcel J. T. Reinders, Ahmed Mahfouz
Summary: scHPL is a hierarchical progressive learning method that can learn cellular hierarchies from multiple datasets while preserving the original annotations.
NATURE COMMUNICATIONS
(2021)
Article
Mathematical & Computational Biology
Jiecong Lin, Xingjian Chen, Ka-Chun Wong
Summary: The issue of off-target cleavage in the CRISPR gene-editing system has been a concern. This study introduces a computational method using convolutional neural network and attention module to predict off-target activity in CRISPR. Validation experiments demonstrate that the proposed model outperforms existing deep-learning-based off-target prediction models in terms of predictive performance.
STATISTICS IN BIOSCIENCES
(2023)
Editorial Material
Computer Science, Artificial Intelligence
Suash Deb, Ka-Chun Wong, Thomas Hanne
NEURAL COMPUTING & APPLICATIONS
(2023)
Article
Computer Science, Artificial Intelligence
Jiejiang Chen, Shaowei Cai, Yiyuan Wang, Wenhao Xu, Jia Ji, Minghao Yin
Summary: The minimum weight dominating set (MWDS) problem is an important generalization of the minimum dominating set problem. In this study, we propose an efficient local search scheme and three novel ideas to improve performance, resulting in the DeepOpt-MWDS algorithm. Extensive experiments show that DeepOpt-MWDS performs better than state-of-the-art algorithms and achieves the best solutions on large-scale graphs.
ARTIFICIAL INTELLIGENCE
(2023)
Article
Biochemical Research Methods
Xiaosa Zhao, Jun Wu, Xiaowei Zhao, Minghao Yin
Summary: This study proposes a new multi-view contrastive heterogeneous graph attention network (GAT) method for predicting lncRNA-disease associations. The method constructs two view graphs using rich biological data sources and designs a cross-contrastive learning task to guide graph embeddings. Experimental results show the effectiveness of this method.
BRIEFINGS IN BIOINFORMATICS
(2023)
Review
Biochemistry & Molecular Biology
Shixiong Zhang, Xiangtao Li, Jiecong Lin, Qiuzhen Lin, Ka-Chun Wong
Summary: The advances in single-cell RNA-seq techniques have allowed for large-scale transcriptomic profiling at single-cell resolution. Unsupervised learning, such as data clustering, is a key component in identifying and characterizing novel cell types and gene expression patterns. This study reviews existing single-cell RNA-seq data clustering methods, including their advantages and limitations, as well as upstream data processing techniques like quality control, normalization, and dimension reduction. Performance comparison experiments evaluate popular single-cell RNA-seq clustering approaches on simulated and multiple single-cell transcriptomic datasets.
Article
Computer Science, Information Systems
Shiwei Pan, Yiming Ma, Yiyuan Wang, Zhiguo Zhou, Jinchao Ji, Minghao Yin, Shuli Hu
Summary: This work presents an improved master-apprentice evolutionary algorithm, MAE-PB, for solving the MIDS problem. The algorithm combines a construction function for generating initial solutions and candidate solution restarting. It uses a multiple neighborhood-based local search algorithm, a recombination strategy based on master and apprentice solutions, and a perturbation strategy for improving solution quality. Computational results on benchmarks and real-world applications demonstrate the high performance of the MAE-PB algorithm.
FRONTIERS OF COMPUTER SCIENCE
(2023)
Article
Biology
Zilong Hou, Yuning Yang, Zhiqiang Ma, Ka-chun Wong, Xiangtao Li
Summary: Protein-protein interactions (PPIs) play a crucial role in cellular pathways and processes, but accurate identification of PPI binding sites is challenging. The proposed EDLM-based method, EDLMPPI, addresses these challenges by utilizing an ensemble deep learning model. Evaluation results demonstrate that EDLMPPI outperforms state-of-the-art techniques in terms of average precision on widely-used benchmark datasets. Additionally, the method provides new insights into protein binding site identification and characterization mechanisms.
COMMUNICATIONS BIOLOGY
(2023)
Article
Computer Science, Artificial Intelligence
Wenhong Wei, Yi Chen, Qiuzhen Lin, Junkai Ji, Ka-Chun Wong, Jianqiang Li
Summary: As people's use of Internet applications increases and concerns about the security of personal data on the Internet grow, cyber security has become increasingly important. Intrusion Detection Systems (IDSs) are crucial tools for detecting and responding to intrusions. Deep Learning (DL) techniques have gained popularity in IDS design due to their promising performance, but their design requires professional knowledge and can significantly impact the DL model's performance. This paper proposes a multi-objective evolutionary DL model (EvoBMF) that incorporates bidirectional Long-short Term Memory (BiLSTM), Multi-Head Attention (MHA), and Full-Connected Layer (FCL) to detect network intrusion behaviors.
APPLIED SOFT COMPUTING
(2023)
Review
Green & Sustainable Science & Technology
Kendric Aaron Tee, Saeed Ahmed, Mohammad A. H. Badsha, Ka Chun James Wong, Irene M. C. Lo
Summary: Due to the strong affinity of lanthanum (La) for phosphate, La compounds such as lanthanum oxide (LO), lanthanum hydroxide (LH), and lanthanum carbonate (LC) have been used in various La-based adsorbents. This study evaluates the differences between LO, LH, and LC in terms of their phosphate removal performance, stability, and reusability. LC has shown superior adsorption capacity, wider pH range, and lower La leaching, making it a potential alternative to LO and LH for phosphate removal. Further studies are needed to compare La compounds in more complex matrices and assess the role of crystal structure in phosphate removal.
CLEAN TECHNOLOGIES AND ENVIRONMENTAL POLICY
(2023)
Article
Computer Science, Interdisciplinary Applications
Rui Sun, Jieyu Wu, Chenghou Jin, Yiyuan Wang, Wenbo Zhou, Minghao Yin
Summary: This paper proposes an efficient local search algorithm based on three main ideas to solve MPIDS problems with different scale instances. The experimental results show that the proposed algorithm performs much better than several state-of-the-art MPIDS algorithms in terms of solution quality.
COMPUTERS & OPERATIONS RESEARCH
(2023)
Article
Computer Science, Interdisciplinary Applications
Jun Wu, Chu-Min Li, Luzhi Wang, Shuli Hu, Peng Zhao, Minghao Yin
Summary: This paper focuses on the problem of finding cohesive groups in a graph, which is important for various real-world applications. The existing methods of searching for cliques are not effective in finding cohesive groups due to their strictness. To address this issue, the Simplified Diversified Top-k s-Plex (S-DTKSP) problem is proposed in this paper. An integer linear programming and an iterated local search algorithm with a tabu strategy are proposed to solve the S-DTKSP problem effectively. Experimental results demonstrate the superiority of the proposed approaches over baseline algorithms.
COMPUTERS & OPERATIONS RESEARCH
(2023)
Article
Computer Science, Artificial Intelligence
Dangdang Niu, Xiaolin Nie, Lilin Zhang, Hongming Zhang, Minghao Yin
Summary: This paper introduces a 0-1 integer linear programming (ILP) model and a framework of greedy randomized adaptive search procedure (GRASP) to solve the minimum weakly connected dominating set problem (MWCDSP). By introducing two novel local search procedures and incorporating greedy functions and a tabu strategy, an improved GRASP algorithm is proposed. Experimental results demonstrate the superior performance of this algorithm over other competitors.
EXPERT SYSTEMS WITH APPLICATIONS
(2023)
Article
Chemistry, Multidisciplinary
Meng Lan, Shixiong Zhang, Lin Gao
Summary: Recent advances in single-cell sequencing technology have led to the development of a deep learning-based framework called scMOG, which can generate single-cell assay for transposase-accessible chromatin (ATAC) data in silico. This framework accurately performs cross-omics generation between RNA and ATAC, and generates paired multiomics data with biological meanings. The generated ATAC data exhibits equivalent or superior performance to that of experimentally measured counterparts. scMOG also proves to be more effective in identifying tumor samples in human lymphoma data than the experimentally measured ATAC data. Moreover, scMOG shows robust performance in generating surface protein data in other omics such as proteomics.
Article
Biochemical Research Methods
Muhammad Toseef, Olutomilayo Olayemi Petinrin, Fuzhou Wang, Saifur Rahaman, Zhe Liu, Xiangtao Li, Ka-Chun Wong
Summary: The use of transfer learning in health informatics and clinical decision-making, particularly in utilizing high-throughput molecular data, has shown great potential in bridging the gap between data domains and overcoming the lack of sufficient training data in clinical research.
BRIEFINGS IN BIOINFORMATICS
(2023)
Article
Computer Science, Artificial Intelligence
Songbai Liu, Qiuzhen Lin, Liang Feng, Ka-Chun Wong, Kay Chen Tan
Summary: Evolutionary transfer optimization (ETO) is a hot topic in evolutionary computation, which seeks to improve optimization efficiency by transferring knowledge across related exercises. This article proposes a multitasking ETO algorithm using transfer learning to solve large-scale multiobjective optimization problems (LMOPs). The algorithm utilizes a discriminative reconstruction network (DRN) for each LMOP to transfer solutions, evaluate correlation, and learn a reduced Pareto-optimal subspace of the target LMOP. The effectiveness of the algorithm is validated in real-world and synthetic problem suites.
IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION
(2023)