Dear all,

a couple of weeks ago I posted a request (below) for a molecular 
dating software suitable to be used programmatically in a sort of pipeline.

Replies (shortened) follow, as some of you requested.

My feeling is that, although some methods exist that are very 
accurate and leave to the operator an ample flexibility in terms of 
analysis design and parameter tweaking, none of these seem to be 
really optimal to work in the context of simulations/database 
automated analyses, where analyses have to be fast, robust and run 
(as much as possible) unsupervised, even at the expense of some 
accuracy/flexibility. Maybe there's some space for improvement here.

..I am not really sure it can do what you want, but have you consider 
the ape package in R? (Miguel)

..MCMCTREE is easy and effective. 
http://abacus.gene.ucl.ac.uk/software/paml.html You can find a step 
by step manual in the end of this website. I think you can implement 
MCMCTREE in your pipeline. You can also implement MULTIDIVTIME in the 
pipeline, but MULTIDIVTIME is little bit old software. (Jun)

..Have you looked at mcmctree in PAML? (Shige)

..You may use DAMBE. The download site also has a set of tutorials. 
The last one (#12, I believe) is on dating. The method is in Xia and 
Yang (2011, MPE). You read in sequences, input a tree, right-click 
one or more nodes and enter the calibration time, set the options, 
and click Run. It is extremely fast and accurate. (Xuhua)

..do you have access to a server (or set of servers)? If so, then you 
could get access to condor and use it to submit and manage the 
execution of jobs. I suggest this because we use condor for this 
purpose on a cluster of computers (94 CPUs) ... As for using BEAST, I 
agree that is recommended. (Lars)

..if you use command line version of beast and write a good 
executable file to run all your files one after the other it will 
need no supervision at all, so maybe this would be a good solution 
for you. (Martyna)

..I suggest to use the software PATHd8 which is very fast, about 1 
second even for very large trees (1000 terminals). The only problem 
is that do not estimate age confidences. http://www2.math.su.se/PATHd8/ (Joan)

Many thanks to everybody,

francesco

-----

Dear all,

I am looking for a molecular dating software that:

- works on a fixed topology and estimates dates/node-heights based on
DNA sequences
- produces a confidence interval for dates/node-heights
- is fast (20 seq, 1kb, 1 minute)

and if possible:

- is reasonably accurate
- allows partitioning of the dataset (if applicable)
- is model based
- can be invoked programmatically under linux

I would generally use BEAST, but this time the analysis has to be
repeated many times and with little supervision,
in a sort of pipeline, hence BEAST would be problematic.

Any help will be appreciated,

francesco

+++++
Francesco Nardi, Dr.

Dept. of Evolutionary Biology
University of Siena

via Aldo Moro 2 - 53100 Siena
Italy

Ph.: +39.0577.234420 (lab. 4398)
Fax.: +39.0577.234476
Web: 
http://www.dbe.unisi.it/index.php?option=com_content&view=article&id=89&Itemid=113&lang=it
+++++

nardifra@unisi.it