Many genomics applications involve examination of signal intensity patterns of probes across the genome, and make inference on the gains and losses of genomic elements from examination of these signal intensities at different chromosome regions. These probes vary greatly in size, ranging from hundreds of kilobases for traditional BAC clone-based array-CGH experiments, to dozens of base pairs for oligonucleotide arrays and high-density single nucleotide polymorphism (SNP) genotyping arrays (1). Typically, a signal intensity measure is calculated for each probe or each probe set, and these intensity values are used to make inference on gains or losses of genomic segments. Various data normalization techniques have been developed to better summarize the intensity values between markers and between experiments, and to accurately capture genomic gains and losses, commonly referred to as copy number variations (CNVs) (2,3).
