MLMeng Liu

Meng Liu

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Meng Liu has a total of 17 co-authors. Here are the top co-authors by number of shared publications:

  • Hong-Yu Ou

    Department of Infectious Diseases, Shanghai Sixth People’s Hospital Affiliated to Shanghai Jiao Tong University School of Medicine , Shanghai 200233

    3 shared publications
  • Cui Tai

    State Key Laboratory of Microbial Metabolism, Joint International Laboratory on Metabolic & Developmental Sciences, School of Life Sciences & Biotechnology, Shanghai Jiao Tong University, Shanghai 200030, China

    2 shared publications
  • Jingyong Sun

    Department of Clinical Microbiology, Ruijin Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai 200025, China

    2 shared publications
  • Xiaobin Li

    State Key Laboratory of Microbial Metabolism, Joint International Laboratory on Metabolic & Developmental Sciences, School of Life Sciences & Biotechnology, Shanghai Jiao Tong University, Shanghai 200030, China

    2 shared publications
  • Yingzhou Xie

    State Key Laboratory of Microbial Metabolism, Joint International Laboratory on Metabolic & Developmental Sciences, School of Life Sciences & Biotechnology, Shanghai Jiao Tong University, Shanghai 200030, China

    2 shared publications
  • Zixin Deng

    State Key Laboratory of Microbial Metabolism, Joint International Laboratory on Metabolic & Developmental Sciences, School of Life Sciences & Biotechnology, Shanghai Jiao Tong University, Shanghai 200030, China

    2 shared publications

Latest Publications

  • EMCLR: Expectation Maximization Contrastive Learning Representations

    by Meng Liu, Ran Yi, Lizhuang Ma

    One of the bottlenecks of self-supervised contrastive learning is the degenerate constant solution, where all the samples are mapped to one single point in representation space. To prevent such collapses, the mainstream paradigm is using negative samples, forcing negative pairs to push away. However, such manner results in ${\mathcal{O}}\left({{N^2}}\right)$ time and space complexities, limiting the expansibility, scalability and efficiency. We observe current negative-requiring objectives can be decomposed to alignment and uniformity, where uniformity dominates the ${\mathcal{O}}\left({{N^2}}\right)$ complexity. To reduce the complexity, inspired by the traditional EM algorithm, we derive the embedding matrix of each batch with optimally uniform distribution and discard the uniformity part in objectives. Specifically, for stacked embedding matrices of two views, we first calculate the optimal solution of one view by the proposed algorithm. Then we align the embedding matrix with the obtained optimal solution. The learning paradigm ingeniously avoids model collapses without ad-hoc negative pairs and reduces the square complexity to linear. Extensive experiments on CIFAR-10/100 and STL-10 show the proposed methods achieve comparable results in ${\mathcal{O}}\left(N\right)$ complexity.

  • Open Access

    ICEO, a biological ontology for representing and analyzing bacterial integrative and conjugative elements

    by Meng Liu, Jialin Liu, Guitian Liu, Hui Wang, Xiaoli Wang, Zixin Deng, Yongqun He, Hong-Yu Ou

    Bacterial integrative and conjugative elements (ICEs) are highly modular mobile genetic elements critical to the horizontal transfer of antibiotic resistance and virulence factor genes. To better understand and analyze the ongoing increase of ICEs, we developed an Integrative and Conjugative Element Ontology (ICEO) to represent the gene components, functional modules, and other information of experimentally verified ICEs. ICEO is aligned with the upper-level Basic Formal Ontology and reuses existing reliable ontologies. There are 31,081 terms, including 26,814 classes from 14 ontologies and 4128 ICEO-specific classes, representing the information of 271 known experimentally verified ICEs from 235 bacterial strains in ICEO currently and 311 predicted ICEs of 272 completely sequenced Klebsiella pneumoniae strains. Three ICEO use cases were illustrated to investigate complex joins of ICEs and their harboring antibiotic resistance or virulence factor genes by using SPARQL or DL query. ICEO has been approved as an Open Biomedical Ontology library ontology. It may be dedicated to facilitating systematical ICE knowledge representation, integration, and computer-assisted queries.


  • Top Cited Papers
    Open Access

    ICEberg 2.0: an updated database of bacterial integrative and conjugative elements

    by Meng Liu, Xiaobin Li, Yingzhou Xie, Dexi Bi, Jingyong Sun, Jun Li, Cui Tai, Zixin Deng, Hong-Yu Ou

    ICEberg 2.0 (http://db-mml.sjtu.edu.cn/ICEberg/) is an updated database that provides comprehensive information about bacterial integrative and conjugative elements (ICEs). Compared with the previous version, three major improvements were made. First, with the aid of text mining and manual curation, it now recorded the details of 1032 ICEs, including 270 with experimental supports and 762 from bioinformatics prediction. Second, as increasing evidence has shown that ICEs frequently mobilize the so-called ‘hitchhikers’, such as integrative and mobilizable elements (IMEs) and cis-mobilizable elements (CIMEs), 83 known transfer interactions between 49 IMEs and 7 CIMEs with 19 ICEs taken from the literature were included and illustrated with visually intuitive directed graphs. An expanded collection of 260 chromosome-borne IMEs and 235 CIMEs was also added. At last, ICEberg 2.0 provides an online tool ICEfinder to predict ICEs or IMEs in bacterial genome sequences. It combines a similarity search for the integrase, relaxase and/or type IV secretion system and the co-localization of these corresponding homologous genes. With the recent updates, ICEberg 2.0 might provide better support for understanding the biological traits of ICEs, especially as their interaction with cognate mobilizable elements may further promote horizontal gene flow.


  • Top Cited Papers
    Open Access

    oriTfinder: a web-based tool for the identification of origin of transfers in DNA sequences of bacterial mobile genetic elements

    by Xiaobin Li, Yingzhou Xie, Meng Liu, Cui Tai, Jingyong Sun, Zixin Deng, Hong-Yu Ou

    OriTfinder is a web server that facilitates the rapid identification of the origin of transfer site (oriT) of a conjugative plasmid or chromosome-borne integrative and conjugative element. The utilized back-end database oriTDB was built upon more than one thousand known oriT regions of bacterial mobile genetic elements (MGEs) as well as the known MGE-encoding relaxases and type IV coupling proteins (T4CP). With a combination of similarity searches for the oriTDB-archived oriT nucleotide sequences and the co-localization of the flanking relaxase homologous genes, the oriTfinder can predict the oriT region with high accuracy in the DNA sequence of a bacterial plasmid or chromosome in minutes. The server also detects the other transfer-related modules, including the potential relaxase gene, T4CP gene and the type IV secretion system gene cluster, and the putative genes coding for virulence factors and acquired antibiotic resistance determinants. oriTfinder may contribute to meeting the increasing demands of re-annotations for bacterial conjugative, mobilizable or non-transferable elements and aid in the rapid risk accession of disease-relevant trait dissemination in pathogenic bacteria of interest. oriTfinder is freely available to all users without any login requirement at http://bioinfo-mml.sjtu.edu.cn/oriTfinder.