Wednesday, October 03, 2007

Phenotype to genotype: as easy as 1,2,3...10   posted by p-ter @ 10/03/2007 07:52:00 PM

I mentioned dog genetics a few days ago; now, a couple papers demonstrating the power of using the dog as a model organism have been published. I don't think people really realize how amazing this is, so I'll try to explain a bit.

Introducing a new model organism to the genetics community is (or perhaps, was) one of those things that happens once a generation or so-- the originals, back at the turn of the century, were the mouse and the fly, pioneered by Morgan and Castle. The organisms were abundant, relatively cheap to maintain, and had generally simple (for what was known at the time) genetics. It's easy to forget that it didn't have to be that way-- if some dude had really liked snakes, maybe the model organism of choice today would be the rattler. But over time, as resources, a community, and a base of knowledge build up, an organism gets entrenched in the system (let's say you're interested in the genetics of wing development-- do you really want to go out butterfly catching when there's a big room full of flies down the hall? And even if you do, what the hell are you going to feed them?).

New organisms have been introduced, of course-- Brenner legendarily elevated the nemotode C. elegans to genetic superstardom, the zebrafish is increasingly popular, and there have always been "niche" organisms like the sea urchin or the subject of this post- the dog.

But until recently, doing dog genetics must have been a real pain. All your friends are knocking out genes, doing fancy mutagenesis screens, and searching genomes for homologous sequence, while you're stuck mapping phenotypes to...oh, somewhere on chromosome 5 (there might be genes in the region, but you don't know, since the genome's not sequenced). Think human genetics circa 1985 (or ok, maybe 1995).

But in the course of a couple years, all that's changed. Utilizing the highly-inbred nature of dog breeds and the recent genome sequence, people were able to develop a SNP chip for genome-wide association using <30,000 SNPs. And with that chip, they found the gene controlling a pigmentation trait in boxers using only 10 (10!) dogs. Estimating the cost of the chips at ~$400 (they're probably less), an interested breeder with a bunch of dogs on his hands could do a decent study the Mendelian phenotype of his choice for less than $20K.

Early mouse geneticists took advantage of the existing infrastructure of mouse fanciers to advance their work; is the dog the new mouse?

Besides humans, the dog is possibly the most interesting organism for taking advantage of new genomic technologies-- the huge variability in phenotype, the massive recent selection to create that variability, and the knowledge of ecology and history all combine to create a gold mine of insight waiting to be tapped.