Re: Fix clock rate in BEAUti 1.8.3

[Andrew Rambaut]

You turn off the operators in the operators panel.

Best,

Andrew

On 30 Mar 2016, at 20:20, Cathy <newm...@gmail.com> wrote:

In BEAUti v.1.8.3, I am trying to set all loci to strict clock with fixed rate=1.0. How do I do this? In BEAST2, there's a checkbox "Estimate" that I un-check for each locus, but that box is missing in 1.8.3, and I don't know what to edit in the XML file. Thanks!

--
You received this message because you are subscribed to the Google Groups "beast-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to beast-users...@googlegroups.com.
To post to this group, send email to beast...@googlegroups.com.
Visit this group at https://groups.google.com/group/beast-users.
For more options, visit https://groups.google.com/d/optout.

[Ryan E.]

Hi Andrew,

Could you provide a more detailed answer to this problem? I'd like to set the substitution rate to 1.55E-8 under the strict clock, and then compare runs where this rate is fixed vs. estimated. I don't see any of these options in BEAUti 1.8.3 that were there previously and it's not clear to me how this could be done using the operators panel. Do I need to modify the xml file?

Thank you,

Ryan

[Andrew Rambaut]

You can set the rate as the initial value in the Priors table (just use uniform prior - it doesn’t matter). Then go to the Operators tab and deselect the operators that alter the rate (there will be 2 - a scale operator on the rate and and up/down operator on the rate and node heights).

Best,

Andrew

[Ryan E.]

Thanks, Andrew. When the strict clock is chosen, there doesn't appear to be an option to enter a rate in the priors (see attached screencap) – that option (a prior for meanRate) only appears for the relaxed and random local clock models. That's not a huge problem because I think I figured out how to enter directly into the xml file, but figured I should let you know.

The more pressing issue issue is that now I keep getting the following repeated message when I change the rate from 1.0 to 1.55E-8 (the same problem occurs with the expanded coalescent model):

SEVERE: CoalescentLikelihood for exponential is -Infinity

Apr 4, 2016 12:50:10 PM dr.evomodel.coalescent.CoalescentLikelihood calculateLogLikelihood

SEVERE: CoalescentLikelihood for exponential is -Infinity

Apr 4, 2016 12:50:10 PM dr.evomodel.coalescent.CoalescentLikelihood calculateLogLikelihood

The rate is in lineage divergence per million years, and I want to use it so that I can compare results between that and a similar analysis that uses an absolute time calibration instead of rate. I found a message board post from 5 years ago (version 1.6.1) where someone had a similar issue and their workaround was to manually set bounds for the growth rate prior to be very close to zero (https://groups.google.com/forum/#!msg/beast-users/_w8nQUelNSo/G8Ny2STqoMsJ). Is this a reasonable thing to do, or could that inappropriately bias my results?

Thanks,

Ryan

[diane.bai...@gmail.com]

Hi everybody.
The same problem here with relaxed clock (in log) and BSP analysis: the ulcd.mean prior never shows up.
It is fixed at 1.0, as mentionned on the buttom.
I go back to Beast v1.6.x, and it's good BUT I'm trying to run Bayesian Skygrid analysis, only available with v1.8.x and I'm sad ...

Remeber to use as well Tracer v1.5 !

[Andrew Rambaut]

Dear Diane,

Try the pre-release of v1.8.4:

https://github.com/beast-dev/beast-mcmc/releases/tag/v1.8.4pre20160319

If that doesn’t fix the issue, let me know or submit an error on the Issue list:

https://github.com/beast-dev/beast-mcmc/issues

Best,

Andrew