Lecture 3

Comparison of Microarray Technologies

October 9, 2001
Lecturer: Larry Ruzzo
Notes: Simon Kahan

Three microarray technologies are available:

Affymetrix GeneChip:

Affymetrix GeneChip technology provides an array of 250,000 probes, each probe containing a set of oligonucleotides of approximately 25 base pairs each representing a region of interest within a gene. As described in the previous lecture, the array is exposed to cDNA developed from a cell in which it is hypothesized these genes are expressed. The cDNA then hybridizes (attaches) to complementary sequences on the array. As with other microarray technologies, the strategy is to identify which genes in the cell are expressed, and to what degree, based on the extent of hybridization observed at each probe. Fluorescent molecules attached to the cDNA create an intensity of light that corresponds to the degree of hybridization.

Typically for each oligo that is intended to exactly match a portion of the gene, Affymetrix ensures there is another oligo that has a mismatch in one location. The purpose of this is to corroborate evidence provided by a match: if one of the probes is heavily hybridized, then the absence of hybridization at the site where the oligo has been deliberately modified supports the hypothesis that the hybridization at the match site was more likely due to a sequence match than to some other source of attraction (eg, that the molecular shape of that oligo and any similar oligo just happens to be that of velcro). The added value of this corroboration was questioned in class.

GeneChips can be used to compare genes expressed in two cells, one of which is known to be afflicted, by comparing the effect of exposing two microarrays to the genetic contents of the cells and comparing the hybridization patterns on those two arrays.

GeneChips are created using a photolithography technology similar to that used in making integrated circuits, though at a 24 micron feature size much larger than the submicron features sizes of ICs. A series of steps proportional in length to the number of bases is required, beginning with a chemically coated glass substrate. A photo-protective mask is used to expose the substrate to light selectively. Where light strikes the subtrate, it becomes active such that with nucleotide incubation, chemical coupling occurs. This must be repeated with different masks for each type of nucleotide, and then repeated again a number of times equal to the length of the oligotide. The whole process is easy to repeat in a mass production of identical GeneChips, but, like IC fabrication, expensive to set up. Creating the masks and carrying through a run costs something like $100,000; each Chip is then sold for about $1000.


Inkjet technology is similar in function. The inkjet controls application of chemicals according to software without the use of physical masks. This makes the setup process much cheaper: there is no need to amortize the cost of a run over a large number of identical cells. However, the probes are less dense than is possible using photolithography.

Spotted Microarrays:

A third technology is the mechanical deposition of entire cDNA or a PCR product onto an array using carefully designed metal pen nibs controlled by a robotic arm.

The array is exposed to equal amounts of green and red fluorescent dyed samples, corresponding to normal and affected cells, respectively. The color at each site indicates the relative amount of hybridization corresponding to the relative expression in the two cells of the cDNA at that site. That is, yellow indicates expression in both cells; black in neither; red in only the affected cell; green only in the normal cell.

Comparison of the Three Technologies:



Spotted Glass

Microarray Business:

In 1997, no one was ready to look at more than 25 to 50 gene levels. Now people want to look at the whole genome. One can imagine using an array per human individual, laced with carefully selected genes representing susceptibility to or contraction of specific diseases, to provide useful diagnostic and preventative medical data. And to establish insurability...

There has been a 45% annual growth in microarray revenues during the past two years. The expected market size is one-billion dollars by 2005.
The Transcriptional Program of Sporulation in Budding Yeast: (an article in Science, Vol 282, October 23, 1998, by S. Chu et al.)
The process by which DNA is involved in the creation of mRNA is called Transcription. Which traits are transcribed changes over the life cycles of a cell; the map of this as a function of cycle state is called a Transcriptional Program. Budding yeast is much like that used in cooking. Sporulation is the life cycle in which spores are produced.
Yeast is normally Diploid, but when producing spores it goes through meiosis producing four haploid daughter cells, which are the spores. It does this when put in a Nitrogen-rich environment as a form of hibernation or self-preservation.
The meiosis is typically described as being divided into four phases: early, middle, mid-late, and late corresponding to replication/recombination, meiosis I, meiosis II, and spore maturation, respectively.
Transcription in the early and middle phases has been studied in experiments predating this paper. 150 genes had been implicated in the sporulation process.
In this work, the authors compare the levels of mRNA present over time using a series of microarrays. They found 500 genes that are induced (ie, whose population increases during the process) and 500 that are repressed.
Additionally, using knockout experiments, they found three specific genes that prevent sporulation but leave other development unchanged.
Graphs show that the number of mononuclear cells drop with time; the number of spores increases; thus demonstrating that the cells they are observing are indeed sporulating. The graphs also demonstrate that it is not a sudden process: the cells are mixed in state over several hours.
Microarray data shows which genes are expressed at the corresponding times. Activity is as expected, given the previous experiments, during the various phases of activity; but clustering the microarray intensities as functions of time into those of similar shape (ie, those probes that experienced a quantity of hybridization that varied similarly over the course of time) suggests division into seven rather than four phases: a refinement of the earlier sporulation process description.

The authors were able, in each of these seven phases, to identify from the microarray data which genes were expressed during that phase; the total being about 500 as compared to 150 known earlier. This gives those who want to better understand the mechanics of the sporulation process a great deal more direction as to where to look for clues.