Hello!
I'm not pretty sure if this is the right blog for posting this, but it seems to be a bug for me.
I'm using TipDate. As result of analysis a substitution rate and confidence interval are shown.
I've used it for at least 4 dataset and always the confidence interval does not included the rate. In fact in order to do that, you have to divide by 1000 the upper and lower values of C.I.
The question is: am I doing something wrong?
Since i have a small datasets I'd used the alingment, and a Exhasutive Tree Search result from PAUP as input tree (lset taked from ModelTest 3.7)

Thank You in advance.

Andrés

Hi Andrés, please would you export the likelihood function to file and either send to me, or paste in here as code. cheers ./w

There is the console output:

---- RUNNING MOLECULAR CLOCK ANALYSIS WITH DATED TIPS----
The underlying methodology developed by Andrew Rambaut is discussed
in Bioinformatics 16(4): 395-399
The following data was read:
6 species:{NOV1998_16301,JUL1999_16524,MAR2000_16772,OCT2001_17356,FEB2002_17490,AGO2006_19107};
Total Sites:594;
Distinct Sites:39
What units are the dates measured in (e.g. months)? This is only used for reporting the results.days
Read the following dates:
Nov1998_16301:      16301      days
Jul1999_16524:      16524      days
Mar2000_16772:      16772      days
Oct2001_17356:      17356      days
Ago2006_19107:      19107      days
Feb2002_17490:      17490      days

CPU time taken: 0.406 seconds.

*-----------------------------------------------------------*
RESULTS WITHOUT THE CLOCK:
Log Likelihood = -1144.61222607557;
Shared Parameters:
CG=0.421889
CT=0.816344
GT=0.160682
AT=0.614661
AC=0.615745

Tree givenTree=(Nov1998_16301:0.00675717,Jul1999_16524:0,(Mar2000_16772:0.0106789,((Oct2001_17356:0.00512315,Ago2006_19107:0.0309416)Node6:0.00497006,Feb2002_17490:0.00604879)Node5:0.00777962)Node3:0.0201417);
*-----------------------------------------------------------*

RESULTS WITH DATED TIPS CLOCK:
Log-likelihood: -1152.64186266955
CPU time taken: 0.922 seconds.

*-----------------------------------------------------------*

-2(Ln Likelihood Ratio)=16.0593
Constrained parameters:4
Asymptotic p-value:0.00294065
Tree with branch lengths scaled in days:
(Nov1998_16301:332.836,Jul1999_16524:555.836,(Mar2000_16772:270.076,((Oct2001_17356:208.657,Ago2006_19107:1959.66):192.178,Feb2002_17490:534.835):453.241):533.761)

Root of the tree placed at 15968.2(95% CI: 15797.7,16089.9) days

Substitution rate estimated at 1.83759e-05(95% CI: 0.138983,0.237546) subs per days per site

======================
Also, I´ve attached 2 comma separated value files with LF (constrained and unconstrainde) (Zipped).

---

Hi again...
Today I'd ran the same analysis (not the posted, but another one) in my home, and at work.
At home, I have installed version 0.992006.....beta of Hyphy for win... but at work I have version 1.0020080508 for Linux. Confidence intervals seems to be right... I mean they included the rate. numericaly are the same, but win version CI values are 10000 times higher than Linux version.
I think that it is an old bug, already fixed.

Thanks Anyway,

PD: nevertheless I get an "out of context" error when I set parameters to be locally estimated.

Hi Andrés,

Glad you sorted the rate issue out - the 0.992006... version is over 3 years old indeed. Also, the analysis can not (because of the underlying model) handle locally estimated parameters. The error you are seeing is not informative (I'll see if this can be improved), but the analysis is not expected to support locally parameterized models.

Sergei

Good!
Thanks a lot folks!!