isa.in.silico {isa2}

Generate in-silico input data for biclustering algorithms

Description

This function generates a test data set for ISA, containing modules of prescribed number, size, signal level, internal noise and background noise.

Usage

isa.in.silico (num.rows = 300, num.cols = 50, num.fact = 3,
     mod.row.size = round(0.5 * num.rows/num.fact), 
     mod.col.size = round(0.5 * num.cols/num.fact), noise = 0.1,                 
     mod.signal = rep(1, num.fact), mod.noise = rep(0, num.fact),                
     overlap.row = 0, overlap.col = overlap.row)

Arguments

num.rows

The number of rows in the data matrix.

num.cols

The number of columns in the data matrix.

num.fact

The number of modules to put into the data.

mod.row.size

The size of the modules, the number of rows per module. It can be a scalar or a vector and it is recycled.

mod.col.size

The size of the modules, the number of columns per module. It can be a scalar or a vector and it is recycled.

noise

The level of the background noise to be added to the data matrix. It gives the standard deviation of the normal distribution from which the noise is generated.

mod.signal

The signal level of the modules.

mod.noise

The noise levels of the different modules. Normally distributed noise with standard deviation mod.noise is added to the data. This is in addition to the background noise.

overlap.row

The overlap of the modules, for the rows. Zero means no overlap, one means one overlapping row, etc.

overlap.col

The overlap of the modules, for the columns. Zero means no overlap, one means one overlapping column, etc.

Details

isa.in.silico creates an artificial data set to test the ISA or any other biclustering algorithm. It creates a data matrix with a checkerboard matrix. In other words, potentially overlapping blocks are planted into a noisy background matrix.

These blocks may have different signal and noise levels and they might also overlap. See the parameters above.

Value

A list with three matrices. The first matrix is the in silico data, the second contains the rows of the correct modules, the third the columns.

Author(s)

Gabor Csardi [e-mail]

References

Bergmann S, Ihmels J, Barkai N: Iterative signature algorithm for the analysis of large-scale gene expression data Phys Rev E Stat Nonlin Soft Matter Phys. 2003 Mar;67(3 Pt 1):031902. Epub 2003 Mar 11.

Ihmels J, Friedlander G, Bergmann S, Sarig O, Ziv Y, Barkai N: Revealing modular organization in the yeast transcriptional network Nat Genet. 2002 Aug;31(4):370-7. Epub 2002 Jul 22

Ihmels J, Bergmann S, Barkai N: Defining transcription modules using large-scale gene expression data Bioinformatics 2004 Sep 1;20(13):1993-2003. Epub 2004 Mar 25.

See Also

isa2-package for a short introduction on the Iterative Signature Algorithm. See isa for an easy way of running ISA.

Examples

## Define a function for plotting if we are interactive
if (interactive()) { layout( rbind(1:2,3:4) ) }
myimage <- function(mat) {
  if (interactive()) { par(mar=c(1,2,2,1)); image(mat[[1]]) }
}

## Create a simple checkerboard without overlap and noise
silico1 <- isa.in.silico(100, 100, 10, mod.row.size=10, mod.col.size=10,
                         noise=0)
myimage(silico1)

## The same, but with some overlap and noise
silico2 <- isa.in.silico(100, 100, 10, mod.row.size=10, mod.col.size=10,
                         noise=0.1, overlap.row=3)
myimage(silico2)

## Modules with different noise levels
silico3 <- isa.in.silico(100, 100, 5, mod.row.size=10, mod.col.size=10,
                         noise=0.01, mod.noise=seq(0.1,by=0.1,length=5))
myimage(silico3)

## Modules with different signal levels
silico4 <- isa.in.silico(100, 100, 5, mod.row.size=10, mod.col.size=10,
                         noise=0.01, mod.signal=seq(1,5,length=5))
myimage(silico4)