fast=1 method which uses a one-sample z-score on the paired LFCs (averaged over samples, then median over inferential replicates). The permutation is computed by changing the signs of the LFC matrix and recomputing z-scores. Testing on the vignette example, but using all the transcripts, the one-sample z-score method takes < 20 seconds while the signed rank method takes > 200 seconds (12x speedup), while they have a high rate of agreement on the detected set (30:1 in common vs discordant).eds package, such that fishpond no longer requires Rcpp and C++ code compilation.fast=1 method which uses a one-sample z-score on the paired LFCs (averaged over samples, then median over inferential replicates). The permutation is computed by changing the signs of the LFC matrix and recomputing z-scores. Testing on the vignette example, but using all the transcripts, the one-sample z-score method takes <20 seconds while the signed rank method takes >200 seconds (12x speedup), while they have a high rate of agreement on the detected set ( 30:1 in common vs discordant).tx2gene argument, so that ranges will be distributed to the rowRanges of the output SummarizedExperiment.shiftX argument to plotInfReps() for numeric x variable, to help with overplotting.Snakemake. See vignette for description of a suggested workflow. For a large single-cell dataset with mean and variance summaries of inferential uncertainty, splitSwish() avoids generating the inferential replicate counts until the data has been split into smaller pieces and sent to different jobs, then only the necessary summary statistics are gathered and q-values computed by addStatsFromCSV().x argument plus grouping with cov for showing counts over pseudotime across groups of cells. Also added applySF argument which can be used to divide out a size factor, and the reorder argument which will re-order the samples/cells within groups by the count. plotInfReps() will draw boxplots with progressively thinner visual features as the number of cells grows to make the plots still legible.