We estimated the power of trendsceek with respect to the identification of genes with an expression hot-spot (A), step-gradient (B), non-radial streak (C) or linear gradients (D). (A-C) Expression values were empirically sampled from seqFISH data. (D) For linear gradients the effect size = highest expression / lowest expression. Sensitivity (y-axis) indicates the proportion of 100 independently simulated genes that was deemed significant (P ≤ 0.05), where cell-positions were randomized for each gene. Vertical lines indicate 95% bootstrap confidence interval of the mean sensitivity from repeated simulations (Methods). (A-D) Sample sizes of 100, 500 and 1000 cells were evaluated with 0.01, 0.05 and 0.1 fraction of spiked cells (A-C).