Fig. 2: Spatial power analysis to recover cells and cellular adjacencies of interest using ISTs in different tissue types.
From: In silico tissue generation and power analysis for spatial omics
a–l, Power analysis of the number and size of FOVs required to detect rare cell types by spatial analysis of tissues with different structures. a–d, HDST data set of breast cancer. a–c, Cells (points) at their spatial position in real (a,b) and corresponding IST (c) data, labeled by type (a) or morphological regions (b). Areas within the red and blue rectangles are expanded in Supplementary Figure 5. d, Probability (y axis) of discovering at least one T cell when sampling different numbers of FOVs (nFOV, x axis) of different sizes (colored lines), in either real tissues (solid lines) or ISTs (dashed lines). e–h, osmFISH murine cortex data. e–g, Cells (points) at their spatial position in real (e,f) and corresponding IST (g) data, labeled by type (e) or morphological region (f). h, Probability (y axis) of discovering at least one L6 pyramidal neuron when sampling different numbers of FOVs (x axis) of different sizes (colored lines), in either real tissues (solid lines) or ISTs (dashed lines). i–l, CODEX mouse spleen data. i–k, Cells (points) at their spatial position in real (e,f) and corresponding IST (g) data, labeled by type as defined by protein expression (i) or morphological region (j). l, Probability (y axis) of discovering at least one megakaryocyte when sampling different numbers of FOVs (x axis) of different sizes (colored lines), in either real tissues (solid lines) or ISTs (dashed lines). d,h,l, Error bars indicate a 95% confidence interval calculated over 50 independent experiments (30 independent FOVs drawn at each size per experiment).
