deconv.entire.genome: High-resolution model-based deconvolution of normalized...

In MeDiChI: MeDiChI ChIP-chip deconvolution library

Description

Deconvolves an entire ChIP-chip data set, across all chromosomes, including multiple replicates, and running multiple bootstraps, if desired, by running 'chip.deconv' on multiple, contiguous, overlapping subsets of the data and combining the results.

Usage

deconv.entire.genome(data, chroms=NA, window=6000, step.by=5500,
centers=NULL, quiet=F, plot=F, n.boot=1, fit.res=10, quant.cutoff="q0.85",
max.peak=NA, verbose=F, kernel=NA, no.multicore=T, ... )

print.chip.deconv.entire.genome(x, ...)

plot.chip.deconv.entire.genome(x, where = NA, center = NA, window = NULL, ...)

coef.chip.deconv.entire.genome(object, ...)

Arguments

data	Input data matrix, connection, or filename. See 'chip.deconv'.
chroms	Character vector listing chromosomes to include in analysis. The default ('NA') is to include all chromosomes. See 'chip.deconv' for more information.
window	Size (in bp) of chunks to be processed one-at-a-time (see 'chip.deconv'). For 'plot.chip.deconv.entire.genome', the coord. range that is plotted; the default ('NULL') is to use the 'window' that was input to 'deconv.entire.genome()'.
step.by	Increment (in bp) in increment the 'center' coordinate of each 'chip.deconv' call.
centers	Optional vector of 'centers' to use (overrides 'step.by' option. Vector may be optionally named by chromosome.

quiet	If TRUE, be very quiet.
plot	If TRUE, plot each chunk as it is deconvolved.
n.boot	Number of bootstrap iterations. See 'chip.deconv'.
fit.res	Deconvolution resolution. See 'chip.deconv'.
quant.cutoff	See 'chip.deconv'.
max.peak	See 'chip.deconv'.
verbose	If TRUE, print out status messages. See 'chip.deconv'.
kernel	Required deconvolution kernel. See 'chip.deconv'.
no.multicore	Prevent use of multiple cores, even if 'multicore' is installed.
x	Object output from 'deconv.entire.genome()'
where	The chromosome(s) to be plotted. See Details.
center	Central chromosomal coordinate(s) to be plotted. See Details.
object	Object output from 'deconv.entire.genome()'
...	Further parameters passed to 'chip.deconv' or 'plot.chip.deconv'.

Details

'deconv.entire.genome' breaks the full data set in to contiguous (in chromosomal coordinate) chunks of size 'window' and processes these using 'chip.deconv'. It increments the center of the 'window' by 'step.by' base pairs, and combines the resulting fit by averaging nearby peaks in the overlapping segments. The defaults were chosen to contain overlaps of about 10%, and constrain the required memory usage for each deconvolution.

'deconv.entire.genome' distributes the processing of multiple sections of the chromosome to multiple processor cores, unless 'no.multicore' is set to TRUE.

'plot.chip.deconv.entire.genome' will, by default, plot the entire fit across all chromosomes, via multiple plots, in chunks of size 'window'. It is recommended to send this to a postscript or pdf device (rather than the screen). See 'plot.chip.deconv' for more information.

Value

A list of class 'chip.deconv.entire.genome', for which 'plot', 'print', and 'coef'. This list is comprised of 'chip.deconv' objects (see 'chip.deconv') for each chromosome. See 'Example' for more information. The list is itemized as:

fits.fin

The fits combined into a single one for each chromosome.

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, plot.chip.deconv, fit.peak.profile, generate.fake.data, generate.binding.profile, MeDiChI-data, lars, quadprog, Matrix

Examples

data( "halo.lowres", package="MeDiChI" )
## Deconvolve all three replicons: "Chr", "pNRC100", "pNRC200". Note
## this will take a while to run.

## Not run: 
fits <- deconv.entire.genome( data.halo.lowres, fit.res=30,
               n.boot=1, kernel=kernel.halo.lowres, verbose=FALSE, trace=FALSE )

## Plot the entire fit, across many windows:
plot( fits )

## Plot a piece of the finished product on just the chromosome:
plot( fits$fits.fin$Chr, center=20000, wind=10000 )

## End(Not run)

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