scaleNormMultCont: Scale multiple conditions

In rhondabacher/SCnorm: Normalization of single cell RNA-seq data

Scale multiple conditions

Description

After conditions are independently normalized with the count-depth effect removed, conditions need to be additionally scaled prior to further analysis. Genes that were normalized in both conditions are split into quartiles based on their un-normalized non-zero medians. Genes in each quartile are scaled to the median fold change of condition specific gene means and overall gene means. This function can be used independetly if SCnorm was run across different Conditions separately. However, the input must be as follow: NormData <- list(list(NormData = normalizedDataSet1), list(NormData = normalizedDataSet2)) where normalizedDataSet1 is the normalized matrix obtained using normcounts() on the output of SCnorm().

Usage

scaleNormMultCont(NormData, OrigData, Genes, useSpikes, useZerosToScale)

Arguments

NormData	list of matrices of normalized expression counts and scale factors for each condition. Matrix rows are genes and columns are samples.
OrigData	list of matrices of un-normalized expression counts. Matrix rows are genes and columns are samples. Each item in list is a different condition.
Genes	vector of genes that will be used to scale conditions, only want to use genes that were normalized.
useSpikes	whether to use spike-ins to perform between condition scaling (default=FALSE). Assumes spike-in names start with "ERCC-".
useZerosToScale	whether to use zeros when scaling across conditions (default=FALSE).

Value

matrix of normalized and scaled expression values for all conditions.

Author(s)

Rhonda Bacher

rhondabacher/SCnorm documentation built on July 8, 2023, 11:36 p.m.