models, clock and skyride results

[Rui Faria]

Dear all,

I have been recently using Beast, specifically the skyride analysis to
inspect demography through time of a fish population based on a mtDNA
gene. I have two main questions:

1. When I choose either TrN or HKY (I did not try with GTR, but maybe
it also holds), but without G+I parameterization, plus a strict
molecular clock, the results are always a flat line with very narrow
confidence intervals. This also happens with example file it comes
with Beast for the skyride analysis. If I choose the same model but
with a relaxed molecular clock (lognormal- 1%), I got some "normal"
oscillations of NE through time Is this result normal, meaning that I
obtain what I usually see for other species. Is there a reason for
this behavior? Or is some kind of bug?

2. In practical terms what is the main difference between extended
bayesian skyline plot (ebsp) and the skyride? are there situations
where we should prefer to use one instead of the other. I'm pretty
convinced by the Mol Bio Evol paper that the smoothing from skyride
works much better on simulated data. I got slightly different results
between ebsp and skyride and I'm not questioning, what makes more
sense to report (although anyway, it is just the demography associated
with a mtDNA gene).

Thanks very much in advance for the feedback.

Best,

Rui

[Rui Faria]

Quarta-feira, 30 de Maio de 2012 18:58:34 UTC+1, Rui Faria escreveu:

Dear all,

I have been recently using Beast, specifically the skyride analysis to
inspect demography through time of a fish population based on a mtDNA
gene. I have two main questions:

1. When I choose either TrN or HKY (I did not try with GTR, but maybe
it also holds), but without G+I parameterization, plus a strict
molecular clock, the results are always a flat line with very narrow
confidence intervals. This also happens with example file it comes
with Beast for the skyride analysis. If I choose the same model but
with a relaxed molecular clock (lognormal- 1%), I got some "normal"
oscillations of NE through time Is this result normal, meaning that I
obtain what I usually see for other species. Is there a reason for
this behavior? Or is some kind of bug?

2. In practical terms what is the main difference between extended
bayesian skyline plot (ebsp) and the skyride? are there situations
where we should prefer to use one instead of the other. I'm pretty
convinced by the Mol Bio Evol paper that the smoothing from skyride
works much better on simulated data. I got slightly different results

between ebsp and skyride and I'm now questioning what makes more

sense to report (although anyway, it is just the demography associated
with a mtDNA gene).

Thanks very much in advance for the feedback.

Best,

Rui