Fig. 1: Power analysis framework for spatial omics data.
From: In silico tissue generation and power analysis for spatial omics
a,b, Features impacting power to detect cell types in single-cell (a) and spatial genomics (b) experiments. c, Use of spatial data sets for retrospective power analysis. Different sizes of FOVs (squares, left) from an existing spatial data set are sampled, and their data (middle) is used to conduct a statistical analysis. The results (right) are used to calculate the probability of detection of a desired feature (y axis, right) when using smaller (orange) or larger (green) FOVs. Dashed line, desired threshold. d, Generation of ISTs. From left: our method generates a blank tissue scaffold using a random-circle-packing algorithm (two left panels), and prior biological knowledge is used to optimize cell-type assignments on the tissue scaffold (second from right), followed by visualization with Voronoi diagrams (right). e, IST generation of complex or large tissues by regional annotations from pilot data. Pilot data (left) are used to assign regional annotations (second left), and spatial parameters are estimated for each region separately. The region-specific parameters are used to generate IST tiles, which are stitched together to create a full IST (second from right), followed by analysis, for example to compare the sampling requirements to detect a spatial feature at a desired power (dashed line) for a small (orange) versus large (green) FOV.
