CHAT (version 1.1)

getOrigin: Inference of Origin Cluster

Description

This function performs K-means on BAF-LRR plot and assign a subset of data points to the Origin Cluster using the concept of consensus clustering.

Usage

getOrigin(sam.dat, ntry = 10, concensus.thr = 0.6, para)

Arguments

sam.dat

numeric matrix, as returned from getSegChr() or getSeg(), with one sample included.

ntry

integer, number of attempts to perform K-means.

concensus.thr

numeric, rate of concensus required to be included in to Origin Cluster. A data point needs to be assigned to the Origin Cluster >= ntry*concensus.thr times to be included.

para

a list of parameters returned from getPara().

Value

a list containing the following elements:

x0

the x coordinate of the Origin Cluster centroid

y0

the y coordinate of the Origin Cluster centroid

list

indices of the segments belong to Origin Cluster in sam.dat.

Details

The Origin Cluster represents the unchanged genome in a tumor sample, and it is important for downstream inference. This function implement a two-step approach to accurately assign the membership to data points. First, it performs concensus clustering on the data points and find a minimum set of the Origin Cluster. It then expands the set by iteratively including more data points into it as long as the new point is close to the centroid of the Cluster.

Examples

para=getPara()

data(A0SD.BAF)
data(A0SD.LRR)
chr8.baf=A0SD.BAF[A0SD.BAF[,2]==8,]
chr8.lrr=A0SD.LRR[A0SD.LRR[,2]==8,]
x=getSegChr(chr8.baf,chr8.lrr)
dd.dat=x[,2:8]
rownames(dd.dat)=x[,1]
mode(dd.dat)='numeric'
oo=getOrigin(dd.dat,para=para)

## See what it looks like on a BAF-LRR plot
plot(dd.dat[,6],dd.dat[,4],pch=19,cex=0.3,xlab='BAF',ylab='LRR',xlim=c(0,0.5),ylim=c(-1,1))
points(x[oo$list,6],dd.dat[oo$list,4],pch=19,cex=0.3,col='yellow')