COO Normalization function

The standard normalization function for RNASeq data

Usage

coo_normalize(cds)

Arguments

cds

- a DESeqDataSet object, with raw counts in the "counts" slot.

Value

A normalized DESeqDataSet

Details

this function applies a robust library size normalization procedure to the raw count matrix provided. The goal is to specifically ensure that the expression values for the data are on a consistent scale, regardless of the source dataset. For more details on the robust library size normalization function, refer to the "COO_Classifier" supplemental vignette.

Please note that rownames of the raw count matrix must be either refseq IDs or ENSEMBL gene IDs. Gene symbols will not work.

Notably, for the purposes of machine learning, we do not care about the absolute expression of genes, but only their expression relative to other samples. This, combined with the fact that people may not know how to get read lengths easily for new data means that we're skipping any normalization step involving read length.