Eric,

Your analysis is very good and largely correct. If you reference the Golub paper, you'll understand why accurate classification is important. The short summary is that the two types of leukemia, ALL and AML, have very similar physiological symptoms. However, while the symptoms appear the same, they are associated with different cells, and the methods of treating them dissimilar. As stated in Golub, "Although remissions [cure] can be achieved using ALL threapy for AML (and vice versa), cure rates are markedly dimished, and unwarranted toxicites are encountered."

Unforunately, in order to correctly classify the two types you have to run all kinds of test and rely on expert judgement. So, a quick and dirty initial test would be ideal in helping with classification (less time waiting, more effective treatment, etc...). That, and a more quantative measure would be nice, as "expert" opinion can be terribly noisy.

There are many examples similar to this one, and if you're interested in this kind of work, the Pharmacology Department here on campus is great. Just ask around about Pharmacogenetics/Pharmacogenomics.

To measure gene expression in a high-throughput way, that is where the DNA microarrays use. Basically, they use lithography techniques to spot a chip with DNA/RNA probes. They extract all the RNA from a cell, and dump it onto the chip. If the RNA binds to a probe, exactly which protein it would express. Of course, that is very watered down and if you want more detail, reference the Shalon paper or talk to Dr. Michael Heller in the Bioengineering department. Lot of active research too, as the technique is very very new.

Little more background, according to the Central Dogma, DNA is converted to mRNA (messenger RNA) in the nucleus, mRNA travels outside the nucleus, where it is translated by the ribosomes which then construct proteins. For more information, see Alberts, the Cell.