isa2 (version 0.3.4)

ppa.normalize: Normalize input data for use with the PPA

Description

Normalize the two input matrices and store them in a form that can be used effectively to perform the Ping-Pong Algorithm

Usage

"ppa.normalize"(data, ...)

Arguments

data
A list of two numeric matrices, with matching number of columns. They might contain Na and/or NaN values.
...
Additional arguments, see details below.

Value

A list of four matrices, the first two corresponds to the first input matrix, the second two to the second matrix.

Details

This function can be called as
    isa.normalize(data, prenormalize = FALSE)
  
where the arguments are:
data
A list of two numeric matrices, with matching number of columns. They might contain Na and/or NaN values.

prenormalize
Logical scalar, see details below.

It was observed that the PPA works best if the input matrices are scaled to have mean zero and standard deviation one.

A PPA step consist of four matrix-multiplications and this requires four different matrices, each of the two input matrices scaled row-wise and column-wise.

If the prenormalized argument is set to TRUE, then row-wise scaling is calculated on the column-wise scaled matrices and not on the raw input.

References

Kutalik Z, Bergmann S, Beckmann, J: A modular approach for integrative analysis of large-scale gene-expression and drug-response data Nat Biotechnol 2008 May; 26(5) 531-9.

See Also

See ppa for an easy way of running the PPA

Examples

Run this code
## In-silico data
set.seed(1)
insili <- ppa.in.silico()
nm <- ppa.normalize(insili[1:2])

## Check it
max(abs(rowSums(nm[[1]])))
max(abs(rowSums(nm[[2]])))
max(abs(rowSums(nm[[3]])))
max(abs(rowSums(nm[[4]])))

max(abs(rowSums(nm[[1]])))
max(abs(rowSums(nm[[2]])))
max(abs(rowSums(nm[[3]])))
max(abs(rowSums(nm[[4]])))

## Plot them
if (interactive()) {
  layout(rbind(1:3,4:6))
  image(insili[[1]], main="Original data 1")
  image(t(nm[[1]]), main="Row normalized 1")
  image(nm[[2]], main="Column normalized 1")

  image(insili[[2]], main="Original data 2")
  image(t(nm[[3]]), main="Row normalized 2")
  image(nm[[4]], main="Column normalized 2")
}

Run the code above in your browser using DataCamp Workspace