Fig. 2: Compound hypothesis testing with PACS is sensitive and specific.
From: Depth-corrected multi-factor dissection of chromatin accessibility for scATAC-seq data with PACS
a–i Comparison of Type I error rates and statistical power across analytical methods using two-factor simulation data. Error bars represent the standard deviation from five independent simulations. All tests conducted are two-sided. The best-performing method in each scenario is highlighted with a color-matched circle. It is important to note that Seurat is limited to testing only cell-type effects. LR denotes logistic regression. j Illustration of linear and quadratic effects of treatment on accessibility across five-time points. The effect sizes are defined as the log fold change (logFC) between the highest and the lowest accessibility values. k Assessment of statistical power for identifying linear and quadratic temporal dynamics across varying effect sizes in simulated datasets. Error bars denote the standard deviation from five independent simulation runs. Superimposed points represent individual simulation power estimates. A sample of N = 1000 cells was analyzed for each time point. l The false positive and false negative rates from ignoring batch covariates in testing cell type effect with ArchR for the adult mouse kidney data.
