An apparatus for gene expression analysis
at the transcriptome
level, gene discovery and/or mutational analysis (depending on type of array).
Aka. (c)DNA/oligonucleotide (micro)array, DNA/oligonucleotide (micro)chip.
Cf. microbead array.
Spots of single-stranded DNA, cDNA or synthesised oligonucleotides (our probes) are deposited on a surface in a regular pattern; we know which DNA is a which position(s). This surface is then incubated with our mRNA or DNA of interest (our targets): probe and target will anneal if they match (sufficiently...), and because we previously attached fluorescent/radioactive tags to our targets (*now* I tell you 8) we can take a CCD image/autoradiogram and note which probes on our surface were present in our target preparation. We can also make estimates of the comparative levels of each target (from the amount of fluorescence/radioactivity), though there are considerable sensitivity issues.
DNA is spotted onto a nylon membrane, with 32/33P used as the targets' tag. These arrays are easy to make, sensitive and reusable. Probe density is low (~5000 spots on a 10cm2 membrane) but sufficient to hold an entire bacterial genome. Cost: ~US$one_grand ea.
cDNA is printed onto microscope slides. Two different target samples are used concurrently, each labelled with a different fluorescent tag to allow for comparative analysis. We can fit thousands of spots on each slide and the manufacture and analysis processes can be automated. However, sensitivity is low (must use duplicates and only differences in expression >2x are meaningful), the microarrays are not reusable, and comparatively large amounts of material are required. Cost: US$cup_of_coffee ea.
Oligonucleotides 20-25 bases long are synthesised on silicon chips, with fluorescent tags on the targets. Sensitivity is worse than with glass microarrays and the chips are single use, but we can detect SNPs, automate the operations, and fit 104s of spots on a single chip! (We can also put oligonucleotides on glass slides; cheaper but lower density.) Note that here we must know the sequence of each probe - no more just dipping into a DNA library - and we must (debatably >;-) own or license the local patents to these sequences...which is why Affymetrix has a virtual monopoly on the business and can charge US$0.4-2_grand ea.