SNP array data offers many advantages in terms of cost, throughput, turn around time and processing. But the key advantage of SNP arrays is the use of SNP BAF in addtion to LRR which really increase the CNV detection sensitivity and a better identification of the event type such as LOH.
SNP array are limited to detecting copy-number differences of sequences present in the reference assembly used to design the probes. Thus it provides no information on the location of duplicated copies. A major limitation of SNP array is the resolution of alterations detected. In order to provide good CNV call a minimum nbumber of probes is required which allows only the detection of large event. Also the resolution of edge of the CNV using SNParray is not effective and would need external data to obtain a base-pair breakpoint.
The most important benefits of NGS technologies are a genome-wide analysis without any prior information, the specificity and linear dynamic range response of NGS data offer many advantages for estimation of copy number. Additionally NGS allows fine detection of small event and several methods of analysis coming from the CGH arrays domain are available. Addtionally CNV breakpoints could often be resolve at the base-pair level.
NGS data are limited by the cost, the turn around time and the processing complexity. Moreover the major limitations of NGS approach for CNV are the use of uniform reads distribution assumption (False) and the inherente noise in the data introduced by the quality of the reference sequence during the alignment step.
When studying CNV, using a combined approach of SNParray and NGS is actually the best alternative.