In certain cases, only a fraction of the samples in a treatment group show a response to the stimulus. To quantify the sensitivity of scCODA in such scenarios, we conducted another benchmark study. We simulated datasets as before, assuming that either a rare or an abundant cell type was increasing by a significant margin in the treatment group (Methods—“Simulation description”). To mimic a partial response to the covariate, we defined treatment groups where the affected cell type was increased in (5%, 10%, … 100%) of the samples, while the rest of the samples followed the distribution of the control group.
