[BioC] LOH analysis for illumina data

[Yu Chuan Tai]

Hi all,

I have some illumina SNP arrays data from a prostate cancer study. Many of 
the tumor samples were contaminated with normal samples such that two 
types of data profiles occurred:

1. LRR are around 0 (2 copies) but BAF has 4 clusters (3 copies)
2. LRR are below 0 (loss) but BAF has 3 clusters (2 copies)

In either case, cnvPartition doesn't estimate the copy numbers correctly.
It called a gain (3 copies) in type 1, and didn't call any loss in type 2.

My question is: Among all the algorithms available there, which one would 
be more suitable for this situation?

Thanks!
Yu Chuan