PASTA: Pattern Analysis for Spatial Omics Data

Author(s): Martin Emons,Samuel Gunz,Helena Lucia Crowell,Mark Robinson

Affiliation(s): University of Zurich



Most spatial omics approaches can be classified under high-throughput sequencing (HTS) based or imaging-based approaches. In HTS-based approaches, positional information is recorded according to the predetermined array of spots measured. Imaging-based approaches, however, either target the molecules of interest with hybridising fluorescent probes, ablate regions stained with a cocktail of antibodies via metal tag readouts, or target sequences are amplified and sequenced in situ. In terms of analysis, the technological streams are distinct: in HTS-based approaches data is collected along the regularly-spaced spots and can be represented as a lattice whereas imaging-based approaches, which are often represented at the cell-level after segmentation and summarisation, can be assumed to be generated by a stochastic process and be represented as a point pattern. In this respect, such modalities have been present in other fields, such as geography, for decades and in particular, the field of (geo)spatial statistics offers many tools for analysis. For example, spatial omics data collected along a regularly-spaced grid can be analysed by lattice data analysis tools, such as Local Moran’s $I$ and Bivariate Lee’s $L$. Imaging data generated by a stochastic process can be input for methods of point pattern analysis, including concepts such as Ripley's $K$ or empty-space functions. In a demo, we will introduce pattern analysis for spatial omics data (PASTA), a toolkit for the exploratory data analysis of spatial omics data. We will highlight the usefulness and transferability of existing spatial data analysis approaches in the context of spatial tissue profiling. Using a vignette that involves data from multiple technologies, concepts will be introduced, assumptions discussed and biological use cases will be shown with inline code. We plan to submit PASTA as a Bioconductor package that seamlessly integrates with existing tools for spatial data analysis such as SpatialExperiment or Voyager.