The commonly used microarray platforms can be categorized into three major types based on the array surfaces and manufacture methods: 1) Affymetrix GeneChip microarrays which contain in situ synthesized 25-mer oligonucleotides (probes) on a glass surface. 2) Microarrays with long DNA sequences printed on a glass slide, where double-stranded (ds) cDNA probes or long oligonucleotide (>50 nt) probes are printed on a chemical treated microscope glass slide. This type of microarray can be manufactured in house or commercially available from Agilent, GE and others. 3) Nylon filter “microarray” is indeed a macroarray where ds cDNA probes or long oligonucleotide (>50 nt) probes are printed on a nylon filter membrane and each probe’s diameter is in the 500 mm range. This type of microarray can also be manufactured in house or available from Clontech and others. It should be kept in mind that each type of microarrays has its own inherent strengths and weaknesses as discussed below. Due to fundamental design differences of various microarray platforms, it is difficult to compare gene expression levels using raw signal values obtained from different platforms. However, using a standardized protocol, a similar gene expression profile could be obtained from different microarray platforms. For details, please see the recent papers on the compatibility and reproducibility among different microarray platforms (microarray related papers).
Affymetrix GeneChip: One of the most commonly used microarray platforms is the Affymetrix GeneChip array because of its robustness and reproducibility. Although custom Affymetrix GeneChip arrays are available, making a new photolithographic mask can be costly. Nevertheless, because of its array design, not only can it be used for differential expression experiments, it can also be used for gene sequencing, the detection of SNPs and other small features in DNA. In general, it has a good quality control and the reproducibility between each Affymetrix GeneChip is greater than 98.5%. Since Affymetrix platform does not employ co-hybridization of control and experimental RNA samples labeled by two different dyes, that is to say that Affymetrix does not produce the ratio of experimental vs. control. In other words, the hybridization of each labeled cRNA produces only an absolute intensity value for each target gene. Hence, it is important to properly normalize the data of control and experimental GeneChips prior to statistical data analysis.
Glass Slide Array: Another very commonly used microarray is a spotting array where ds cDNA or long oligonucleotides (>50 nt) are printed on a chemical treated glass microscope slide. Since this type of array can be manufactured in house, it is relatively inexpensive and very flexible in array design. Especially glass slide microarray platform employs co-hybridizations of control and experimental RNA samples labeled by two different dyes (Cy3 and Cy5), the control and experimental targets can be co-hybridized on a single microarray at the same time versus on two Affymetrix GeneChips separately with a single dye (streptavidin phycoerythrin staining). However, the spotting process itself is inherently variable and the reproducibility among arrays greatly depends on the quality of glass slides. It should be kept in mind that there are couple technical problems to employ ds cDNAs as probes. For example, because of the length of long ds cDNA (> 200 bp) probes, it is difficult to discern among genes with more than 80% of homology nor can it detect SNPs. Nevertheless, since the price for long oligonucleotide synthesis has decreased, it is possible to overcome these limitations by spotting long oligos on glass slides.
Nylon Filter Array: The manufacture of a nylon filter array is similar to that of a glass slide array in using ds cDNA or long oligos as probes. The targets prepared for the hybridization can be labeled with florescent dye or radioisotope (usually P33) but the latter can be unfavorable to some researchers. Nevertheless, the hybridization of radioisotope-labeled targets produces high quality data which cover signal intensities within a linear range of over 105. Since nylon filter platform does not employ co-hybridization of control and experimental RNA samples labeled by two different dyes, each hybridization can only produce an absolute intensity value for each gene similar to Affymetrix GeneChip platform. It is also important to properly normalize the data from control and experimental nylon filter arrays prior to statistical data analysis.
