[SS21] Transformation and Integration of Data and Modeling in Medicine

[SS21-07] High-order optimization method for multi-modal data integration

*Hao Jiang¹, Wai-Ki Ching², Dong Shen¹ (1. Renmin University of China (China), 2. The University of Hong Kong (Hong Kong))

Keywords:

single cell,multi-omics,optimization,high-order

Simultaneous profiling of multi-omics single-cell data provides deeper insights into cellular states and functions. Cellular Indexing of Transcriptomes and Epitopes by Sequencing (CITE-seq) enables the parallel quantification of cell-surface protein expression and transcriptome profiling within the same cells. Similarly, Methylome and transcriptome sequencing from single cells (scM&T-Seq) facilitates the analysis of transcriptomic and epigenomic profiling in individual cells. However, there is an increasing demand for effective integration methods to accurately capture the heterogeneity of cells from noisy, sparse, and complex multi-modal data. We will address the problem of heterogeneity analysis and representation learning in multi-modal data, with the development of high-order laplacian matrix optimization method.