To test the algorithm, our results were compared to the mapping obtained using other cytogenetics and/or molecular genetics technologies. We showed that our method is able to produce accurate copy number detection and high-resolution breakpoint identification. The advantages of our approach are presented and discussed in comparison to the only other current software, BeadStudio LOH+ (Illumina). We believe the OB-HMM method is highly suited to the analysis of high-throughput genomic data when one of the hidden states has special status as a ‘null’ or normal state. In this case, the OB-HMM allows for setting of parameters which ensure certain frequentist coverage properties for excursions of the model out of the null state, while benefiting from Bayesian marginal inference. To our knowledge, we are the first to consider OB-HMM for genomic data analysis, and we believe the framework we have developed is well suited to many other genomic data types, including other SNP array platforms and array CGH. In previous work, several other authors have considered conventional HMM-based statistical methods to detect copy number changes using array CGH (23,24) and GeneChip®
