ABSTRACT Single-cell omics assays have become essential tools for identifying and characterizing cell types and states of complex tissues. While each single-modality assay reveals distinctive features about the sequenced cells, true multi-omics assays are still in early stage of development. This notion signifies the importance of computationally integrating single-cell omics data that are conducted on various samples across various modalities. In addition, the advent of multiplexed molecular imaging assays has given rise to a need for computational methods for integrative analysis of single-cell imaging and omics data. Here, we present GLUER (inte G rative ana L ysis of m U lti-omics at single-c E ll R esolution), a flexible tool for integration of single-cell multi-omics data and imaging data. Using multiple true multi-omics data sets as the ground truth, we demonstrate that GLUER achieved significant improvement over existing methods in terms of the accuracy of matching cells across different data modalities resulting in ameliorating downstream analyses such as clustering and trajectory inference. We further demonstrate the broad utility of GLUER for integrating single-cell transcriptomics data with imaging-based spatial proteomics and transcriptomics data. Finally, we extend GLUER to leverage true cell-pair labels when available in true multi-omics data, and show that this approach improves co-embedding and clustering results. With the rapid accumulation of single-cell multi-omics and imaging data, integrated data holds the promise of furthering our understanding of the role of heterogeneity in development and disease.

GLUER: integrative analysis of single-cell omics and imaging data by deep neural network