biomvRCNS (version 1.12.0)

regionSegCost: Regional segmentation cost matrix

Description

To calculate regional cost matrix for the initial stage and second merging stage of the segmentation model.

Usage

regionSegCost(x, maxk = NULL, segs = NULL, family = NULL, alpha = NULL, useSum = TRUE, useMC = FALSE, comVar = TRUE)

Arguments

x

The input data matrix or vector

maxk

Maximum number of index to search forward

segs

Starting indices (excluding 1) for the candidate segments, for the second stage model, maxk will be overridden with length(segs)+1.

family

which exponential family the data belongs to, possible values are 'norm', 'pois' and 'nbinom'

alpha

alpha matrix for negative binomial cost calculation, estimated from regionSegAlphaNB

useSum

TRUE if using grand sum across sample / x columns, like in the tilingArray solution

useMC

TRUE if mclapply should be used to speed up

comVar

TRUE if assuming common variance across samples (x columns)

Value

Matrix with maxk rows and nrow(x) columns, or a length(segs)+1 square matrix for the second stage model.

Details

Preparing the cost matrix for the follow-up segmentation. Using residual sum of squares for 'norm' data, and negative log-likelihood for 'pois' and 'nbinom' data. Extension of the costMatrix function in tilingArray.

References

Piegorsch, W. W. (1990). Maximum likelihood estimation for the negative binomial dispersion parameter. Biometrics, 863-867.

Picard,F. et al. (2005) A statistical approach for array CGH data analysis. BMC Bioinformatics, 6, 27. Huber,W. et al. (2006) Transcript mapping with high density oligonucleotide tiling arrays. Bioinformatics, 22, 1963-1970. .

Robinson MD and Smyth GK (2008). Small-sample estimation of negative binomial dispersion, with applications to SAGE data. Biostatistics, 9, 321-332

See Also

regionSegAlphaNB

Examples

	x<-matrix(rnorm(120), ncol=3)
	Ca<-regionSegCost(x, maxk=20, family='norm')
	dim(Ca) # [1] 20 40
	Cb<-regionSegCost(x, segs=as.integer(c(3, 6, 12, 30)), family='norm')
	dim(Cb) # [1] 5 5