generate.fake.data: Generate simulated ChIP-chip data.

In MeDiChI: MeDiChI ChIP-chip deconvolution library

Description

Generate simulated ChIP-chip data, as described in Reference (1).

Usage

generate.fake.data(posns = seq(1, 6001, by = 20), n.pts = 5, noise =
0.1, min.pk = 0.1, plot = F, verbose = F, in.pts = NULL, kernel = NULL,
tile.size = 100, reps = 1,
noise.func = function(data)noise*(1+sqrt(data)), ...)

Arguments

posns	Probe centers
n.pts	Number of (randomly-chosen) binding sites to add
noise	Level of noise (as described in Reference (1))
min.pk	Minimum peak intensity
plot	Plot the generated data?
verbose	Be verbose?
in.pts	2-column matrix with positions (column 1) and intensities (column 2) of input peaks; if n.pts is 'NA'.
kernel	Input peak profile to use.
tile.size	Input probe length (used to generate peak profile kernel if 'kernel' is 'NULL')
noise.func	Function used to generate noise as a function of signal intensity
reps	Number of replicate intensities per simulated probe
...	Further parameters for 'generate.binding.profile' (if 'kernel' is 'NULL')

Details

No details.

Value

A list of class containing the following three elements:

input	A two-column matrix containing positions and intensities of input binding sites
data	A two-column matrix containing positions and intensities of simulated probes, can be passed directly to 'chip.deconv'
kernel	The peak profile kernel used to generate the data (as produced by 'generate.binding.profile')

Author(s)

David J Reiss, Institute for Systems Biology

Maintainer: <[e-mail]>

References

Reiss, DJ and Facciotti, MT and Baliga, NS. (2007). "Model-based deconvolution of genome-wide DNA binding", Bioinformatics; doi: 10.1093/bioinformatics/btm592.
http://baliga.systemsbiology.net/medichi

See Also

chip.deconv, generate.binding.profile

Examples

## Generate data with 2 peaks at positions 3000 and 4000, deconvolve the
## data, and plot the resulting data and fit.
kern.300 <- generate.binding.profile( fragment=function(x) dgamma( x,
                     shape=6, scale=50 ), verbose=TRUE )
data <- generate.fake.data( in.pts=cbind( c( 3000, 4000 ), c( 1, 0.7 ) ), reps=3, 
                     kernel=kern.300 )
plot( data$data, pch=20 )
fit <- chip.deconv( data$data, center=NA, wind=NA, kernel=data$kernel,
                    fit.res=30, n.boot=10, verbose=TRUE, boot.sample.opt="case" )
plot( fit, boot="prob.scaled" )
print( fit )

MeDiChI documentation built on May 2, 2019, 5:32 p.m.