R: Removes the effect of the cell-cycle

ccRemover {ccRemover}

R Documentation

Removes the effect of the cell-cycle

Description

ccRemover returns a data matrix with the effects of the cell-cycle removed.

Usage

ccRemover(dat, cutoff = 3, max_it = 4, nboot = 200, ntop = 10,
  bar = TRUE)

Arguments

`dat`	A list containing a data frame , `x`, that contains gene expression measurements with each column representing a sample and each row representing a gene and a logical vector, `if_cc`, that indicates which of the genes/rows are related to the cell-cycle or factor of interest. It is recommended that the elements of x are log-transformed and centered for each gene. For example if `x` contains TPM measurements then we suggest the following two-steps: Step 1: `dat$x <- log(dat$x + 1)` Step 2: `dat$x` - rowMeans(dat$x) ccRemover requires that the samples have been properly normalized for sequencing depth and we recommend doing so prior to applying the above steps. The `if_cc` vector must be the same length as the number of rows in `x` and have elements equal to `TRUE` for genes which are related to the cell-cycle and and elements equal to `FALSE` for genes which are unrelated to the cell-cycle.
`cutoff`	The significance cutoff for identifying sources of variation related to the cell-cycle. The default value is 3, which roughly corresponds to a p-value of 0.01.
`max_it`	The maximum number of iterations for the algorithm. The default value is 4.
`nboot`	The number of bootstrap repititions to be carried out on each iteration to determine the significance of each component.
`ntop`	The number of components considered tested at each iteration as cell-cycle effects. The default value if 10
`bar`	Whether to display a progress bar or not. The progress bar will not work in R-markdown enviornments so this option may be turned off. The default value is `TRUE`.

Details

Implements the algorithm described in Barron, M. & Li, J. "Identifying and removing the cell-cycle effect from scRNA-Sequencing data" (2016), Scientific Reports. This function takes a normalized, log-transformed and centered matrix of scRNA-seq expression data and a list of genes which are known to be related to the cell-cycle effect. It then captures the main sources of variation in the data and determines which of these are related to the cell-cycle before removing those that are. Please see the original manuscript for further details.

Value

A data matrix with the effects of the cell-cycle removed.

Examples

set.seed(10)
# Load in example data
data(t.cell_data)
head(t.cell_data[,1:5])
# Center data and select small sample for example
t_cell_data_cen <- t(scale(t(t.cell_data[,1:20]), center=TRUE, scale=FALSE))
# Extract gene names
gene_names <- rownames(t_cell_data_cen)
# Determine which genes are annotated to the cell-cycle
cell_cycle_gene_indices <- gene_indexer(gene_names,
species = "mouse", name_type = "symbol")
# Create "if_cc" vector
if_cc <- rep(FALSE,nrow(t_cell_data_cen))
if_cc[cell_cycle_gene_indices] <- TRUE
# Move data into list
dat <- list(x=t_cell_data_cen, if_cc=if_cc)
# Run ccRemover
## Not run: 
 xhat <- ccRemover(dat, cutoff = 3, max_it = 4, nboot = 200, ntop = 10)

## End(Not run)
# Run ccRemover with reduced bootstrap repetitions for example only
xhat <- ccRemover(dat, cutoff = 3, max_it = 4, nboot = 20, ntop = 10)
head(xhat[,1:5])
# Run ccRemover with more compoents considered
## Not run: 
xhat <- ccRemover(dat, cutoff = 3, max_it = 4, nboot = 200, ntop = 15)
 
## End(Not run)

[Package ccRemover version 1.0.4 Index]