autoSPIN: A wrapper function for autoSPIN sorting method

In JinmiaoChenLab/uSORT: uSORT: A self-refining ordering pipeline for gene selection

Description

A wrapper function for autoSPIN method which implements optimized local refinement using the selected SPIN sorting method, i.e. STS or Neighborhood.

Usage

autoSPIN(data, data_type = c("linear", "cyclical"),
  sorting_method = c("STS", "neighborhood"), alpha = 0.2, sigma_width = 1,
  no_randomization = 20, window_perc_range = c(0.1, 0.9),
  window_size_incre_perct = 0.05)

Arguments

data	An log2 transformed expresssion matrix containing n-rows of cells and m-cols of genes.
data_type	A character string indicating the type of progression, i.e. 'linear' (strictly linear) or 'cyclical' (cyclically linear).
sorting_method	A character string indicating the choice of SPIN sorting method, i.e. 'STS' (Side-to-Side) or 'Neighborhood'.
alpha	A fraction value denoting the size of locality used for calculating the summed local variance.
sigma_width	An integer number denoting the degree of spread of the gaussian distribution which is used for computing weight matrix for Neighborhood sorting method.
no_randomization	An integer number indicating the number of repeated sorting, each of which uses randomly selected initial cell position.
window_perc_range	A fraction value indicating the range of window size to be examined during local refinement.
window_size_incre_perct	A fraction value indicating the step size at each iteration for incrementing window size.

Value

A data frame containing single column of ordered sample IDs.

Examples

set.seed(15)
da <- iris[sample(150, 150, replace = FALSE), ]
rownames(da) <- paste0('spl_',seq(1,nrow(da)))
d <- da[,1:4]
dl <- da[,5,drop=FALSE]
res <- autoSPIN(data = d)
dl <- dl[match(res$SampleID,rownames(dl)),]
annot <- data.frame(id = seq(1,nrow(res)), label=dl, stringsAsFactors = FALSE)
#ggplot(annot, aes(x=id, y=id, colour = label)) + geom_point() + theme_bw()

JinmiaoChenLab/uSORT documentation built on May 7, 2019, 10:53 a.m.