Evaluation of FDR and power for QuadST across different parameters and data sets, and comparison of its performance with alternative methods in various settings. (A) FDR and power of QuadST using different number of quantile levels. (B) FDR and power of QuadST using different number of k nearest neighbors (KNN) at k = 1, …, 5. (C) FDR and power of QuadST in data with different cell numbers per cell type. (D) Comparison of different methods in an ideal case where measurement error and confounding factor do not exist and interacting cell pairs needed for Giotto and NCEM are perfectly known. (E) Comparison of different methods where interacting cell pairs needed for Giotto and NCEM cannot be perfectly identified. (F) Comparison of different methods under the existence of measurement error for cell–cell distance. (G) Comparison of different methods under the existence of unobtained confounding factor.