The genes are allocated to a bin based on their average log-level of expression, then with each bin the dispersion (variance over mean of the log-levels) is z-scored. Genes having a sufficiently high dispersion z-scores are returned. This method replicates FindVariableFeatures from the Seurat package.

Arguments

zscore_threshold

a numeric value indicating the zcored dispersion threshold above which the gene names are returned. Default to 0.

num_bins

a integer value indicating the number of bins used to calculate z-score into. Default to 20.

data_status

character string. Specifies whether the gene expression levels used for calculation are raw ("Raw"), normalized ("Normalized") or have been imputed ("Smoothed"). Default to "Raw".

invert

logical. If FALSE (default) genes with z-scored dispersion higher than the threshold are returned. If FALSE, the complementary set is returned.

Value

a character vector of genes showing a sufficient level of dispersion.