Efficiently Indexing Cellular Spatial Locations by Importing Multi-layer Regionalization

We propose Polyomino, which uniquely leverages spatial region information to index the spatial localization of cells. Polyomino demonstrates robustness in integrating data with various types of sequencing noise, such as cell segmentation errors and disproportionate cell-type representation.Polyomino was 10x to 362x faster than state-of-the-art (SOTA) methods, and it is the only method that can integrate datasets more than 0.5 million cells in a single run.