calibration with fossil date

[Anthony]

Hi guys, just a basic question, I am trying to calibrate an analysis
using a fossil date (45 million years). I picked the lognormal prior
in BEAUti, defined the mean (45) and std dev (0.05) and ticked the
'mean in real space box', which I found out the hard way was critical.
But, can someone tell me please what the 'Offset' is and how to set
it? Also, by putting 45 into the log(mean) spot, it does not seem to
really fix the lower bound too tightly, which I thought was the idea.
Am i doing something wrong? Thanks for any feedback.

[Luca Pozzi]

Hi Anthony,

the offset defines your "minimum bound" when you calibrate a node. So
in your case, if you want to fix the lower/minimum bound at 45Ma you
should set the offset at 45 and then play with mean and std dev in
order to get a 95% confidence interval that makes sense according to
your prior knowledge of the fossil record. In any case with a
lognormal distribution you will have a soft upper/maximum bound. If
you set the mean at 45 and offset=0 you are basically using also a
lower soft bound instead of a hard constraint.

For a much more detailed explanation you can read this post on the
forum:
http://groups.google.com/group/beast-users/browse_thread/thread/1dde79dfaab20b19

Hope this can help.

Luca

[Anthony]

Thank you very much Luca,

I have tried manipulating the mean and stdev values by directly
changing the xml code, e.g.,

<logNormalPrior mean="2.0" stdev="0.5" offset="45.0"
meanInRealSpace="true">
<statistic idref="tmrca(fossil group)"/>
</logNormalPrior>

but get an error. does anyone have any idea please?

cheers,

Anthony

On Jul 2, 2:33 am, Luca Pozzi <lucapozz...@gmail.com> wrote:
> Hi Anthony,
>
> the offset defines your "minimum bound" when you calibrate a node. So
> in your case, if you want to fix the lower/minimum bound at 45Ma you
> should set the offset at 45 and then play with mean and std dev in
> order to get a 95% confidence interval that makes sense according to
> your prior knowledge of the fossil record. In any case with a
> lognormal distribution you will have a soft upper/maximum bound. If
> you set the mean at 45 and offset=0 you are basically using also a
> lower soft bound instead of a hard constraint.
>
> For a much more detailed explanation you can read this post on the

> forum:http://groups.google.com/group/beast-users/browse_thread/thread/1dde7...

[Gerald Schneeweiss]

Hi,

the mean cannot be younger (less) than the offset, as the offset defines a
minimum age.

Best,
Gerald

> --
> You received this message because you are subscribed to the Google Groups
> "beast-users" group.
> To post to this group, send email to beast...@googlegroups.com.
> To unsubscribe from this group, send email to
> beast-users...@googlegroups.com.
> For more options, visit this group at
> http://groups.google.com/group/beast-users?hl=en.
>
>
>

[Sere Bix]

Well, to narrow it down we would need to know what kind of error message you get.

All the best,
Sere