If you could use next-gen sequencing technology to answer any question....

[ECO]

...what small study would you do?

This is intended as a bit of fantasy, but a serious question, as I've often considered this question:

If you could do one small set of focused samples what would they be and what platform would you use to answer the question?

Cost/labor/your ability to analyze the data is not to be considered...we're talking pure scientific curiousity. Let's limit it to say one run on a GA/SOLiD, or 2-3 runs on 454, or a combination of short/long reads. But no more than 8-16 samples.

Moon rocks? Environmental mining? The fauna scraped off your windshield after a cross country drive? Swabs from a subway seat?

[cariaso]

N timepoints for 3 patients from birth to death. Patients are triplets (twins+1) . The first two would be raised in the same environment, the third would be separated at birth. Sequence the epigenome. No clear platform preference.

No triplets were harmed in the making of this fantasy.

[apfejes]

I like cariaso's study, though I'd also do gene expression studies:

I'd take N drug therapies, and test the whole transcriptome shotgun expression on each of M cell types, on X people, where N, N and X are impossibly large numbers. And, of course, each of the X people would also have a complete genome reference assembled.

You'd be able to figure out what parts of the genome cause drug interactions (successful, no effect or adverse side effect) if M,N and X were large enough.

For people who develop resistance, ie. cancer drugs no longer work, I'd also WTSS those tissues before and after resistance occurs to identify mechanisms.

What fun!