cdnormbio

Condition-Decomposition Normalization for Biological Applications

cdnormbio is an R package that implements Condition-Decomposition methods for data normalization, such as MedianCD normalization and SVCD normalization, for the analysis of data from high-throughput biology assays.

For more details please see:
Roca C.P., Gomes S.I.L., Amorim M.J.B. & Scott-Fordsmand J.J. Variation-preserving normalization unveils blind spots in gene expression profiling. Sci. Rep. 7, 42460; doi:10.1038/srep42460 (2017).

Installation

To install cdnormbio from this GitHub repository, use the function install_github in the devtools package.

library( devtools )
install_github( "carlosproca/cdnormbio" )

Help

Use the standard help in R.

library( cdnormbio )

? normalize.svcd
? normalize.mediancd

Examples

A minimal example:

library( cdnormbio )

gene.n <- 1000
sample.n <- 9
expr.data <- matrix( rnorm( gene.n * sample.n ), nrow = gene.n )
expr.condition <- rep( c( 1, 2, 3 ), each = 3 )

normalize.result <- normalize.svcd( expr.data, expr.condition )

expr.data.normalized <- normalize.result$data

Another, a bit more complex, example:

library( cdnormbio )

gene.n <- 10000
sample.n <- 9
expr.data <- matrix( rnorm( gene.n * sample.n ), nrow = gene.n )
expr.condition <- rep( c( "treatment.a", "treatment.b", "control" ), each = 3 )
offset.added <- rnorm( sample.n )
expr.data <- sweep( expr.data, 2, offset.added, "+" )

normalize.result <- normalize.svcd( expr.data, expr.condition, 
    stdvec.graph = "stdvec", p.value.graph = "p_value", verbose = TRUE )

expr.data.normalized <- normalize.result$data

boxplot( expr.data )
boxplot( expr.data.normalized )

normalization.offset <- normalize.result$offset
sd( normalize.result$offset - offset.added )

normalization.h0.feature <- normalize.result$h0.feature
length( normalize.result$h0.feature )