Normalizations: Profile Normalization Methods

In tallulandrews/TreeOfCells: Inferring the Tree of Cells

Description

Methods for normalizing cell-type profiles.

Usage

   quantile_norm(merged_profiles, conditions=NULL)
   scnorm(merged_profiles, conditions=rep(1, ncol(merged_profiles$mus)))
   scaled_regression(merged_profiles, conditions=NULL)
   sctrans(merged_profiles, conditions=NULL)
   log_transform(merged_profiles, conditions=NULL)
   total_norm(merged_profiles, min_rs = 10^-10, conditions=NULL)
   gene_length_norm(merged_profiles, gene_lengths, conditions=NULL)

Arguments

merged_profiles	Output from: merge_profiles function.
conditions	vector of different condition (only used in scnorm).
gene_lengths	a data.frame with two columns: symbol, length. The former is the gene IDs that matches your cell-type profiles.
min_rs	scalar value, minimum allowable value for the dispersion parameter "r"

Details

Normalizes the various parameters of the cell-type profiles using different methods:

scaled_regression: uses the nlme package, to perform weighted linear regression to eliminate the effect of total expression from each gene independently. Then scales each gene to have a mean of 0 and sd of 1.

quantile_norm: quantile normalizes all parameters using preprocessCore::normalize.quantiles scnorm: normalizes all parameters using SCnorm. sctrans: normalizes means using sctransform. log_transform: log (base e) transforms both means and dispersion parameters. total_norm: normalizes both means and dispersion parameters using the median total expression. gene_length_norm: rescales each gene by the provided gene lengths.

Value

merged_profiles object with corrected matrices.

Examples

obj <- list(
	mus = matrix(rnorm(200), ncol=10),
	rs  = matrix(rgamma(200, shape=0.75, scale=1), ncol=10),
	ds  = matrix(runif(200), ncol=10),
	Ns  = matrix(rpois(200, lambda=30), ncol=10))
colnames(obj$mus) <- paste("celltype", 1:ncol(obj$mus), sep="")
rownames(obj$mus) <- paste("gene", 1:nrow(obj$mus), sep="")
n_obj1 <- quantile_norm(obj)
n_obj2 <- scaled_regression(obj)
n_obj3 <- total_norm(obj)
n_obj4 <- log_transform(obj)
n_obj5 <- gene_length_norm(obj, gene_lengths=list(symbol=rownames(obj$mus), length=1:20))

tallulandrews/TreeOfCells documentation built on April 26, 2020, 2:43 p.m.