I’ll answer everyone’s questions at once. First, I’d like to emphasize that I’ve studied this technique, but I’ve never actually done it, so feel free to enlighten me if I’ve gotten anything wrong.
Okay. So you’re a molecular biologist, and you want to study the expression of genes under different conditions. It’s not so hard to study the change in expression of one gene - take some samples, do reverse-transcription (turn all the RNA back to DNA), run a PCR, and compare the levels of expression in both. But what if you want to study the change in expression of every gene in the human body - all 30,000? You’d want to do that, for instance, to see the difference in protein expression in muscle versus skin tissue, or in bacteria grown with or without glucose. Say you’ve identified the pathway involved in glucose metabolism, but you want to see which other genes are affected by the presence of glucose. For the molecular biologists of a generation ago, you would have had to do the whole thing by hand - 30,000 RT-PCRs, or 30,000 Northern blots - the task basically was impossible.
Now along comes this brand new technique, the microarray, which can be highly automated and lets you study a gargantuan amount of information. First, you take a sample of the RNA from a group of cells from each condition - let’s say condition A and condition B. You reverse-transcribe the DNA, but you fluorescently label each base used for polymerization. For example, you’d label all of your condition A DNA red, and all of your condition B DNA green. On a microarray chip, you’ve attached a bunch of probes - single stranded copies of a portion of a gene (let’s say gene 1), which will bind to the DNA you’ve prepared. Then you mix the condition A and condition B DNA on the chip. Whichever sample has a higher concentration of gene 1 DNA will hybridize more with the probe, and stick to the microarray chip. So, if the cells in condition A express gene 1 more than the cells in condition B, the chip will fluoresce red. If both genes are expressed at the same rate, the result is usually expressed as a yellow dot. Therefore, the color of the chip tells you under which condition you have more gene expression. If you’re confused, this illustration might help.
So that’s how you would do a microarray for one gene. But why stop at just one? You’re dealing with probes the width of a few atoms, and you can program a computer to analyze your results. That means you can massively scale up your data collection. Here’s a picture of a microarray of 40,000 genes. At that point, the limiting factor in your research is how much time you can dedicate to analyzing all the raw data you receive. But on the plus side, you’re left with a mesmerizing microarray chip.
Thanks for all the interest! I’m currently taking an awesome class on analytical techniques (for chemistry - I learned about microarrays in another class), so I’ve been introduced to a ton of cool techniques no one has ever heard of. I’ll try to periodically write about some of the cooler ones, but feel free to ask me anything about sweet science techniques.