Testing natural selection with genetics
H. Allen Orr (one of the authors of the study I mentioned recently) has a fun article in Scientific American on testing hypotheses about natural selection using genetic data. Orr has been one of the few people to try and formally model adaptation in a population genetics framework (I highly recommend this review article from 2005 for a well-written and accessible discussion of this issue), so his thoughts are worth a read.
And if you’re in the mood for a chuckle, check out Larry Moran’s thoughts on the article.
Labels: Genetics
i found out some weird things recently, though i guess it shouldn’t have surprised me
1) “larry moran” is actually a perl script designed to pop out particular outputs based on a set of inputs (e.g., r. a. fisher -> motoo kimura!!!! motoo kimura!!!)
2) “alex vargas” is a newer proptoype, coded in python.
oh, an stephen jay gould was actually really old cobol code that was finally ditched a few years back when no new programmers could be found who were willing to maintain the spaghetii….
If you google “invoke / invoking selection” vs. “invoke / invoking drift,” the straw-maning of selection is an order of magnitude more common. Ditto for searching JSTOR.
“We don’t need to invoke selection…” — yeah, and we don’t need to invoke drift either. There are only three possible cases: the data you have are consistent with neither drift nor selection, with both, or with just one of them. See which case it is, and be done with it. Then maybe tabulate which case it is at various loci, and do some kind of statistical test — just like that selectionist boogeyman John Gillespie does in The Causes of Molecular Evolution, where he tests whether the Poissonian molecular clock fits or doesn’t fit the data across a large number of loci.
JSTOR shows the “invoke selection” swear word became popular in the late ’70s and afterward, although it’s rare but present back to 1957. So this funny debate Larry Moran’s in is obviously an effect of grad school cohort — if you were born around 1945 – 1950, you and your age-mates will be deeply divided on the “selection or drift” non-debate, where one side caricatures the other, which doesn’t bother responding.
it actually doesn’t bother me when people talk about not having to “invoke selection”. In some sense, a selective model is like a neutral model with an additional parameter. So it feels natural to have to justify including an additional parameter in a model (like in all hypothesis testing). though it’s true it’s not exactly analogous.
Now see, I tend to think of “invoke” in terms like “Surrounded by Level 9 Idiots, the Wizard invoked his Dark Incantation, banishing his puny Foes to the Land of Eternal Drift.”
But that’s just me.