Well, it seems like, as with most technologies, it depends on the question you are asking.
The paper listed below, published in Genome Biology (http://genomebiology.com/2011/12/9/R95/abstract),
Comprehensive comparison of three commercial human whole-exome captureplatforms
Asan2,3,*, Yu Xu1,*, Hui Jiang1,*, Chris Tyler-Smith4,*, Yali Xue4, Tao Jiang1, JiaweiWang1, Mingzhi Wu1, Xiao Liu1, Geng Tian1, Jun Wang1, Jian Wang1, HuangmingYang1,# and Xiuqing Zhang1,#.
compared similar measurements across 1) NibleGen Sequence Capture Array, 2) NimbleGen SeqCap EZ and 3) Agilent SureSelect technologies.
The authors used similar preparation procedures for the three technologies – same number of PCR cycles and minimizing the DNA input, reducing the likelihood that external variables would affect the results.
It is comforting that all three technologies performed similarly with respect to calling SNPs, assigning genotypes (>77% sensitivity), and reproducing results. All three also exhibited similar GC and reference allele biases (which I suppose is good, although it might have been nice if one had much lower biases). Both technologies were biased against low (<20%) or high (>75%) GC content, but the array technology actually performed a little better coverage of low GC-content regions. However, all technologies performed very poorly, missing most high GC (>72-80%) content regions.
As far as differences:
-Agilent covered approximately 1000 more genes and 100 more microRNA genes than NimbleGen(4.1Mb of sequence specific to NimbleGen and 7.6Mb specific to Agilent). Still, 15,883 genes overlapped between the two technologies. I looked through the supplemental materials and couldn’t find out whether the differences were due to differential captures of sex-linked genes (X or Y).
– NimbleGen provided better flanking-sequence capture (although this technology is designed to capture coding sequences, often 200bp flanking each side may also be captured)
– NimbleGen also had more uniform coverage and greater genotype sensitivity at high sequencing depth (30-100X)
A drawback of exome sequences is that they represent less than two percent of the entire genome, so functionally relevant changes in regulatory elements, or other non-coding elements will be missed.
Alternatively, some advantages of exome sequencing are that it focuses the search to the parts of the genome that we know are functionally relevant, and these regions are more likely to be involved in human diseases than a randomly selected non-coding region.
In summary, all three technologies perform similarly well, but one should be aware that Agilent has the best gene coverage, SeqCap EZ captures more flanking sequences and the one array capture (NimbleGen Sequence Capture) performs better with respect to covering low GC (<20%) regions.