Fig. 3

Sample size estimated by Sensei. a Estimated sample size for detecting statistically significant difference in normal tissue and primary tumor using an one-sided Welch’s t-test at a significance level of 0.05 with 80% power (the same below). Estimations for unpaired test and paired test are shown in blue and yellow, respectively. Estimations are for infinite (the legacy approach, left end of a whisker), 1,000 (left bar), 384 (right bar, may overlap with the left one), and 100 (right end of a whisker) cells. Fewer cells per sample would require more samples to ascertain an effect. The estimated sample size is for each of the two group in a controlled study, not jointly. For matched-pairs study, it is the same as the number of participants. Sample sizes larger than 200 are omitted. The direction of change in cell type abundance is shown by an arrow. An up arrow indicates a higher abundance in primary tumor compared with normal tissue, and vice versa. b Estimated sample size for detecting statistically significant difference in primary tumor and recurrent tumor for low grade glioma (LGG) and glioblastoma multiforme (GBM) patients. An up arrow indicates a higher abundance in recurrent tumor compared with primary tumor, and vice versa. c Estimated sample size for detecting statistically significant difference in each immune cell type between microsatellite instability-high (MSI-H) and microsatellite stable (MSS) tumor samples in uterine corpus endometrial carcinoma (UCEC), colon adenocarcinoma (COAD), and stomach adenocarcinoma (STAD). An up arrow indicates a higher abundance in MSI-H tumor compared with MSS tumor, and vice versa. d Estimated sample size for detecting statistically significant difference between pre- and post- treatment samples from metastatic melanoma patients. An up arrow indicates a higher abundance in post-treatment tumor compared with pre-treatment tumor, and vice versa.