RJaCGH (version 1.5.7)

pREC_S: Subgroups of arrays that share common alterations

Description

An algorithm to find regions of gain/lost copy number shared by a given proportion of arrays over a probability threshold.

Usage

pREC_S(obj, p, freq.array, alteration = "Gain",
       force.write.files = FALSE, verbose = FALSE,
       delete.rewritten = TRUE)

Arguments

obj

An object of class 'RJaCGH.array'.

p

Threshold for the minimum joint probability of the region on every array.

freq.array

Minimum number of arrays that share every region.

alteration

Either 'Gain' or 'Loss'.

force.write.files

If TRUE, the gzipped files with the Viterbi sequence are written to disk even if they are, supposedly, already there.

verbose

If TRUE provide more details on what is being done, including intermediate output from the C functions themselves. Only helpful for debugging or if you are bored; this will often write more output than you want.

delete.rewritten

If TRUE, delete the files with the Viterbi sequence from disk.

Value

• An object of class pREC_S.none, or pREC_S.Chromosomes depending on whether or not the original RJaCGH had, or not, a Chromosomes component (the later only when the original object was of neither "RJaCGH.Chrom" or "RJaCGH.Genome"). They are lists with a sublist for every region encountered and elements:
• start
  Start position of the region.
• indexStart
  index position of the start of the region.
• indexEnd
  index position of the end of the region.
• end
  End position of the region.
• members
  Arrays that share the region.
• If there are chromosome information, this information will be enclosed in a list for each chromosome.

Details

This algorithm, as pREC_A computes probabilistic common regions but instead of finding regions that have a joint probability of alteration over all arrays, pREC_S searches for regions that have a probability of alteration higher than a threshold in at least a minimum number of arrays. So, pREC_S finds subsets of arrays that share subsets of alterations. Please note that if the method returns several sets or regions, the probability of alteration of all of them doesn't have to be over the probability threshold; in other words p is computed for every region, not for all the sequence of regions.

Writing the files with the Vitterbi sequence to disk will be done by default if RJaCGH was run with the default "delete_gzipped = TRUE" (which will happen if the global flag ".__DELETE_GZIPPED" is TRUE). To preserve disk space, as soon as the gzipped files are read into R (and incorporated into the RJaCGH object), by default they are deleted from disk. Sometimes, however, you might want not to delete them from disk; for instance, if you will continue working from this directory, and you want to save some CPU time. If the files exist in the directory when you call pREC there is no need to write them from R to disk, which allows you to save the time in the "writeBin" calls inside pREC. In this case, you would run RJaCGH with "delete_gzipped = FALSE". Now, if for some reason those files are no longer available (you move directories, you delete them, etc), you should set "force.write.files = TRUE" (pREC will let you know if you need to do so). delete.rewritten helps prevent cluttering the disk. Files with the Viterbi sequence will be written to disk, read by C, and then deleted. Again, no information is lost, since the sequences are stored as part of the RJaCGH object. Note, however, that if you run RJaCGH with ".__DELETE_GZIPPED <- FALSE" this option has no effect, because it is implicit that you wanted, from the start, to preserve the files. In other words, "delete.rewritten" only has any effect if you either used "force.write.files = TRUE" or if you originally run RJaCGH without preserving the files in disk.

References

Rueda OM, Diaz-Uriarte R. Flexible and Accurate Detection of Genomic Copy-Number Changes from aCGH. PLoS Comput Biol. 2007;3(6):e122

See Also

RJaCGH, states, model.averaging, print.pREC_S plot.pREC_S pREC_A

Examples

## MCR for a single array:
y <- c(rnorm(100, 0, 1), rnorm(10, -3, 1), rnorm(20, 3, 1),
       rnorm(100,0, 1)) 
z <- c(rnorm(110, 0, 1), rnorm(20, 3, 1),
       rnorm(100,0, 1)) 
zz <- c(rnorm(90, 0, 1), rnorm(40, 3, 1),
       rnorm(100,0, 1)) 

Pos <- sample(x=1:500, size=230, replace=TRUE)
Pos <- cumsum(Pos)
Chrom <- rep(1:23, rep(10, 23))

jp <- list(sigma.tau.mu=rep(0.05, 4), sigma.tau.sigma.2=rep(0.03, 4),
           sigma.tau.beta=rep(0.07, 4), tau.split.mu=0.1)

fit.array.genome <- RJaCGH(y=cbind(y,z,zz),
Pos=Pos, Chrom=Chrom, model="Genome",
burnin=1000, TOT=1000, jump.parameters=jp, k.max = 4)
pREC_S(fit.array.genome, p=0.4, freq.array=2,
alteration="Gain")
pREC_S(fit.array.genome, p=0.4, freq.array=2, alteration="Loss")