GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2009-02 > 1235691680


From: "Tim Janzen" <>
Subject: Re: [DNA] Puzzle, resolved?
Date: Thu, 26 Feb 2009 15:42:57 -0800
In-Reply-To: <49A71586.4020004@ucl.ac.uk>


Dear James,
Thanks for this interesting information and results. It seems to me
that the population growth rate is an important variable that hasn't been
consistently incorporated into TMRCA estimates.
Does anyone know how to factor the population growth rate into a
TRMCA estimate using the variance method? Does anyone know of published
estimates of population growth rates during the past 10,000 to 20,000 years?
There is some information that pertains to this at
http://en.wikipedia.org/wiki/World_population.
Sincerely,
Tim Janzen

-----Original Message-----
From:
[mailto:] On Behalf Of James Heald
Sent: Thursday, February 26, 2009 2:20 PM
To:
Subject: Re: [DNA] Puzzle, resolved?


Rather than numbers of generations, internally Batwing uses and reports
a "scaled time", T , measured in numbers of generations divided by N,
the variable for the final population size. At least, that's what the
manual leads one to believe.

In fact what it actually does, I think, is to scale T by the *initial*
population size, /not/ the final size. If you do this, suddenly all the
shapes and numbers make sense, whereas otherwise they don't. It doesn't
make any difference for model 0 (constant population size), and I
haven't tried model 2 (break from constant to exponential). But if
you're analysing the data in R and using Batwing's model 1 (exponential
growth), I think you need a line like this to convert the output from
Batwing into a vector of samples of times in generations:

o <- read.table("out",header=T) # read in the data
o <- o[-c(1:1000),] # drop the first 1000 samples
t1 <- o[,"T"] * o[,"N"] # extract T and N and multiply them
alpha <- o[,"alpha"] # extract growth rate alpha
t2 <- t1 * exp(-alpha * t1) # adjust N to N_initial


But it seems to me Batwing is well worth a play.

-- James.

This thread: