Re: tmrca distribution not normal

[pepster]

Have you seen figure 2 in "Calibrated Tree Priors for Relaxed Phylogenetics and Divergence Time Estimation" (http://sysbio.oxfordjournals.org/content/61/1/138.short)?

-Joseph

On Tuesday, March 26, 2013 7:25:30 AM UTC+13, Sophie wrote:

Hi,

I am new with Beast/Beauty 1.7.5. I want to infer the age of 2 nodes using a calibration time I have on a 3rd node (the root between 8-12 Mya) on the COI gene. I manage to get good ESS values for all my parameters using a lognormal tmrca with an offset of 8, playing with mean (10) and log(stdev)(3.7) to have a 95% at 12. However, the output I got for my 2 estimated nodes are a bit weird (see pictures). I got a "normal like distribution" however there is an "artifactual peak" that shows up (picture 21 and 23). This "artifactual peak" showed up exactly at the position of the median I calibrated. I try increasing the chain lenght and changing a bit the mean (2) and log(stdev) (0.5) so the prior distribution look closer to a normal distribution, but the "peak" got even wider (pictures 23, and 24). HKY seems to better estimate the ESS. I've tried also 2 and 3 partitions into codon position but got the same thing. I attached you the parameters I used as well.

What am I doing wrong? 

Thanks in advance for your help

Sophie

[Sophie]

Thanks! It does look like that. I've tried the monophyly option but didn't seem to fix the problem (but I'll try again since I made so many runs now I am not sure if I've tried with the same parameters). I used a random starting tree and the "Speciation Yule process". If I understand correctly, using the "Speciation: calibrated Yule" and a prior tree might help. Is that right?

[Sophie]

Is there a tutorial/documentation available for how to use the "Speciation: calibrated Yule" tree prior with Beauty? I cannot find any and I get an error message "calibrated Yule requieres 1 calibrated internal node with a proper prior and monophyly enforced for each tree". I checked the "monophyly" box for my nodes, so I guess the problem is probably in how I gave the priors.

[Peter]

Hi, Sophie and BEAST community:
I meet the same issues.  Could you share your solution with me?
Thanks,

All the best,

Runhua

[Cajsa Lisa Anderson]

Same problem here... I can't find a way around the "calibrated Yule requieres 1 calibrated internal node with a proper prior and monophyly enforced for each tree".
Anyone found the solution...?

Cheers,
Cajsa