discretize {fcoex} | R Documentation |
Set the discretized expression attribute Uses the discretize_exprs function of the FCBF package
discretize( fc, number_of_bins = 4, method = "varying_width", alpha = 1, centers = 3, min_max_cutoff = 0.25, show_pb = TRUE ) ## S4 method for signature 'fcoex' discretize( fc, number_of_bins = 4, method = "varying_width", alpha = 1, centers = 3, min_max_cutoff = 0.25, show_pb = TRUE )
fc |
Object of class |
number_of_bins |
Number of equal-width bins for discretization. Note: it is a binary discretization, with the first bin becoming one class ('low') and the other bins, another class ('high').#' Defaults to 4. |
method |
Method applied to all genes for discretization. Methods available: "varying_width" (Binarization modulated by the number_of_bins param), "mean" (Split in ON/OFF by each gene mean expression), "median" (Split in ON/OFF by each gene median expression), "mean_sd"(Split in low/medium/high by each assigning "medium" to the interval between mean +- standard_deviation. Modulated by the alpha param, which enlarges (>1) or shrinks (<1) the "medium" interval. ), ), "kmeans"(Split in different groups by the kmeans algorithm. As many groups as specified by the centers param) and "min_max_%" (Similat to the "varying width", a binarization threshold in a "GMM" (A Gaussian Mixture Model as implemented by the package mclust, trying to fit 2:5 Gaussians). Default is "varying_width" |
alpha |
Modulator for the "mean_sd" method.Enlarges (>1) or shrinks (<1) the "medium" interval. Defaults to 1. |
centers |
Modulator for the "kmeans" method. Defaults to 3. |
min_max_cutoff |
<- Modulator for the "min_max_%" method. Defaults to 0.25. |
show_pb |
Enables a progress bar for the discretization. Defaults to TRUE. |
A data frame with the discretized features in the same order as previously
library(SingleCellExperiment) data("mini_pbmc3k") targets <- colData(mini_pbmc3k)$clusters exprs <- as.data.frame(assay(mini_pbmc3k, "logcounts")) fc <- new_fcoex(exprs, targets) fc <- discretize(fc)