A Novel CNA/LOH Detection Algorithm Using Normal-Tumor SNP-Array Data

Recently single nucleotide polymorphism (SNP) genotyping arrays attracts lots of attentions, which can provide high resolution profiling chromosomal rearrangements. It facilitates whole genome detection of two common aberrations: copy number alteration (CNA) and loss of heterozygosity (LOH), which are frequently found in cancer cells. At present, many computational approaches have been introduced for this purpose, however, most of them fail to incorporate the intrinsic genetic relationship between tumor and paired normal SNP-array data, which may greatly improve the performance in identifying CNA and LOH in cancer genome. To address this issue, we proposed a novel algorithm to handle paired SNP-array data from both tumor and paired normal samples, which can make best use of the genotype information to assist the detection. This algorithm employs the statistical framework of HMM and EM method to precisely model the relationship between normal and tumor SNP-array data. Results on public datasets show that our method outperforms all other investigated algorithms with precise parameter estimation and sensitive aberration identification.