metaseqR (version 1.12.2)

normalize.noiseq: Normalization based on the NOISeq package

Description

This function is a wrapper over NOISeq normalization. It accepts a matrix of gene counts (e.g. produced by importing an externally generated table of counts to the main metaseqr pipeline).

Usage

normalize.noiseq(gene.counts, sample.list,
        norm.args = NULL, gene.data = NULL, log.offset = 1,
        output = c("matrix", "native"))

Arguments

gene.counts

a table where each row represents a gene and each column a sample. Each cell contains the read counts for each gene and sample. Such a table can be produced outside metaseqr and is imported during the basic metaseqr workflow.

sample.list

the list containing condition names and the samples under each condition.

norm.args

a list of NOISeq normalization parameters. See the result of get.defaults("normalization", "noiseq") for an example and how you can modify it.

gene.data

an optional annotation data frame (such the ones produced by get.annotation which contains the GC content for each gene and from which the gene lengths can be inferred by chromosome coordinates.

log.offset

an offset to use to avoid infinity in logarithmic data transformations.

output

the class of the output object. It can be "matrix" (default) for versatility with other tools or "native" for the NOISeq native S4 object (SeqExpressionSet). In the latter case it should be handled with suitable NOISeq methods.

Value

• A matrix with normalized counts.

Examples

require(DESeq)
data.matrix <- counts(makeExampleCountDataSet())
sample.list <- list(A=c("A1","A2"),B=c("B1","B2","B3"))
diagplot.boxplot(data.matrix,sample.list)

lengths <- round(1000*runif(nrow(data.matrix)))
starts <- round(1000*runif(nrow(data.matrix)))
ends <- starts + lengths
gc=runif(nrow(data.matrix))
gene.data <- data.frame(
    chromosome=c(rep("chr1",nrow(data.matrix)/2),
        rep("chr2",nrow(data.matrix)/2)),
    start=starts,end=ends,gene_id=rownames(data.matrix),gc_content=gc
)
norm.data.matrix <- normalize.noiseq(data.matrix,sample.list,gene.data)
diagplot.boxplot(norm.data.matrix,sample.list)