Multiscale Data Integration for Single-Cell Spatial Genomics

Technological advances in measuring gene expression in a spatially resolved manner have resulted in several publicly available datasets, often accompanied by sample-matched dissociated single-cell RNA-seq or single-cell multi-omic measurements. However, methodologies for analyzing such data are in urgent need of development. Currently, many integrative data analysis tasks for spatial genomics are performed using tools designed with dissociated single-cell RNA-seq data in mind, effectively ignoring the specific data structures of spatial genomics data.

This project will develop novel methodologies for multiscale data integration of single-cell spatial genomics in three key ways:

• Developing a within-dataset data integration approach for single-cell spatial genomics to address the confounding of biological and technical factors across spatial coordinates;
• Designing and assessing multiscale (subcellular, cellular and supercellular) features to perform mosaic data integration across spatial genomics and dissociated multiomics datasets; and
• Extending methods for spatial reconstruction of dissociated single-cell data by harnessing multiple spatial references and enabling use of non-shared features among spatial references coming from multiple technologies.

This project will result in novel methodologies and software towards performing multiscale data integration of spatial genomics data. The tools developed during this project will enable the building of fully harmonized spatial and non-spatial data resources.