Whose technology works better? At the Macroarray Results through Microarrays Conference in Boston last week, two posters dared to compare different commercially available technologies for specific experimental applications.
Researchers at the Mayo Clinic’s translational genomics center had a technology problem on their hands when they were called upon to analyze samples in an association study between polymorphisms and diabetic neuropathy that presented complex demands: genotyping a SNP, a pentanucleotide repeat, a biallelic 27 base pair variable number tandem repeat (VNTR), and an insertion/deletion polymorphism.
The researchers, led by Youvraj Sohni, solved this problem with a combination of technologies. They found that the Nanogen NanoChip platform, which consists of 100 individually addressable electrodes that anchor down DNA or oligonucleotide probes in a workstation, worked best for assaying the SNP.
For the other polymorphisms, the group used the Agilent Bioanalyzer with the Caliper DNA 500 LabChip. This system enabled them to size fragments of DNA using a sizing ladder that consisted of a mixture of pre-sized identified DNA fragments. They confirmed the amplicon identities using GeneScan software.
“The application and use of several technologies may be essential for pharmacogenomic and gene-disease association studies,” the researchers concluded.
Agilent vs. Affy
To get their heads around the problem of which technology works better for expression profiling of human brain RNA, researchers at Psychiatric Genomics in Gaithersburg, Md., decided to run a head-to-head comparison of Affymetrix and Agilent microarrays.
The group hybridized the same set of samples, from normal, bipolar, and schizophrenic human brains, and a control rat frontal cortex sample to five Affymetrix 21,000-gene arrays, and six Agilent 12,000-gene arrays. To analyze the data quality, they used scatter plots comparing self to self, control to bipolar, and control to schizophrenic. Both showed low noise levels and little skewing of intensities.
The difference showed up when it came to differentially expressed genes. The Affymetrix chips detected 4.3 percent of the genes as differentially expressed, and the Agilent chips detected 2.3 percent of the genes, with a p-value of .001. To see how much these profiles overlapped, the researchers randomly chose 10 differentially expressed genes on the Agilent array. They found that eight were also differentially expressed on the Affymetrix array.
They also found that the Agilent arrays needed six replicates to eliminate enough false positives to reach a p-value less than .05, while five Affymetrix arrays were needed.
The coefficient of variation on the Agilent arrays was .731 and the CV on the Affymetrix arrays was .706 when the Absent-Absent calls were eliminated.
In all, the researchers found that while the Affymetrix arrays offered a greater quantity of information, the Agilent arrays offered a comparable quality of data. “The Agilent and Affymetrix platforms performed well for gene expression profiling of human postmortem brain tissue, even if the RNA is less than optimal quality,” they wrote at the conclusion of their poster.