CHAT (version 1.1)

getSegPurity: Segment-specific AGP inference by sample

Description

This function implements the sAGP inference algorithm, by placing each of the data point onto a BAF-LRR plot. AGP inference must be done in prior.

Usage

getSegPurity(seg.dat, oo, AGP = 1, type = 1, para, rm.thr = 50, ref.dd = NULL)

Arguments

seg.dat

numeric matrix, segmentation data returned from getSegChr() or getSeg()

oo

list, origin information returned from getOrigin() or getDiploidOrigin()

AGP

numeric, AGP value for the sample being evaluated

type

integer, ploidy type of tumors. 1, diploid; 2, tetraploid; 3, hexaploid.

para

list, parameters returned fro getPara.sAGP()

rm.thr

integer, segments with number of BAF markers below this threshold will be removed.

ref.dd

numeric matrix, in the same format as sam.dd. If given, no copy number estimation will be performed, and only sAGP will be inferred.

Value

a numeric matrix with following columns: chromosome, start position, end position, LRR value, number of LRR markers, BAF value, number of BAF markers, sAGP, number of minor alleles, number of total alleles.

Details

For a data point A on BAF-LRR plot, the algorithm search through all the canonical lines and find the nearest ones to A. Each canonical line corresponds to the contraction path from a canonical point with (nb, nt) towards the origin, where nb is the number of minor allele and nt the number of total alleles. It then uses an empirical objective function F=nt-2*type+K*|sAGP-AGP|, to determine which line to choose. sAGP is the estimated aneuploid fraction of the specific segment for a given canonical line and K the constant parameter set in para. The purporse of this function is to find the most parsimonious combination of nb, nt and sAGP, with parsimony meaning close to genome-wide average.

Examples

data(A0SD.BAF)
data(A0SD.LRR)
seg.dat=c()
for(CHR in c(8,9,10)){
	baf=A0SD.BAF[A0SD.BAF[,2]==CHR,]
	lrr=A0SD.LRR[A0SD.LRR[,2]==CHR,]
	x=getSegChr(baf,lrr)
	seg.dat=rbind(seg.dat,x)
}
dd.dat=seg.dat[,2:8]
rownames(dd.dat)=seg.dat[,1]
mode(dd.dat)='numeric'
para.s=getPara.sAGP()
para=getPara()
oo=getOrigin(dd.dat,para=para)
sAGP.dat=getSegPurity(dd.dat,oo,AGP=0.5,para=para.s)