1 September 2023 in IEEE Journal of Biomedical and Health Informatics
It is critical to correctly assemble high-dimensional single-cell RNA sequencing (scRNA-seq) datasets and downscale them for downstream analysis. However, given the complex relationships between cells, it remains a challenge to simultaneously eliminate batch ef-fects between datasets and maintain the topology between cells within each dataset. Here, we propose scGAMNN, a deep learning model based on graph autoencoder, to simultaneously achieve batch correction and topology-preserving dimensionality reduction. The low-dimensional integrated data obtained by scGAMNN can be used for visu-alization, clustering and trajectory inference.By comparing it with the other five methods, multiple tasks show that scGAMNN consistently has comparable data integration performance in clustering and trajectory conservation.
7 April 2025 in IEEE Transactions on Computational Biology and Bioinformatics
Currently, microbial drug resistance has become a serious global problem. Investigating the relationship between microbes and drugs can facilitate understanding of microbial resistance to certain drugs and help develop more effective therapeutic strategies. Therefore, an efficient and rapid computational method for identifying microbe-drug associations is very important for optimizing antimicrobial therapy, monitoring drug resistance, and addressing bacterial resistance challenges. In this paper, a model called ANFAISMDA for identifying potential microbe-drug associations is proposed. First, Structural information of microbes and drugs is obtained based on 16S rRNA gene sequences of microbes and SMILES structures of drugs, respectively. Topological information of microbes and drugs is obtained based on symmetric matrix completion algorithm, respectively. Then, the structural and topological information is effectively fused according to our proposed adaptive network fusion algorithm. The algorithm combines network node and path information while also adaptively updating weights based on inter-network transmissibility. Finally, an integration strategy is proposed to significantly improve model performance and thereby more accurately identify novel microbe-drug associations. Overall, The reliability and validity of the model are fully verified by experiments. Further, we also have visualized the top 50 microbe-drug associations and found some interesting association patterns. The results of the study show the importance of our approach in optimizing antimicrobial therapy, monitoring drug resistance, and addressing the challenges of bacterial resistance. In addition, the study confirms the usefulness of the model in gaining a deeper understanding of the complex mechanisms of microbe-drug associations.