Hi all!
I've my sequences, viral ones, serially taken from a patient. I make the best posible tree (model test, then paup exhaustive search). I ran the DateTipsMolecularClock script, and get some results... a LRT between noclock and clock hypotheses, and a rate with confidence intervals
The question is:
If I get a p value for the test of, say, 0,025... Is the "molecular clock hypothesis" barely rejected? how is constructed the LRT? and last but not least... What does the rates mean if I rejected the clock hypotesis?

Thank you very much!

Regards!

Andrés

Yes...
The estimations of beast are 10 times higher. However I´m working with very small datasets (6-10 sequences of 700 bp) and I´ve read somewhere that Felsestein said that coalescent analysis are OK for as les as 8 sequences if they are long enough (not my case)

The first approach of my analysis was the evaluation of direct sequences of the viruses (coinfected host). The sequences seem to contain temporal data (topology of tree is compatible with the sampling, and the branch lengths are strongly correlated with the time) but since correlation assumptions are violated by definition by the phylogenies, I was looking for some "quick and dirt, yet better" way for estimating rates... Dated tips sounded fine for me, but then... there is no evidence of clock.

I´ve also tried beast, and in fact the "real" rate analyses were carried out with cloned sequences (about 18 for sample) with beast. The rates I get with beast are x10 or x100 the results of dated tips.

As I understand now, neither the Date Tips nor the beast estimations are reliable since for the former there is not a strict clock and for the last the sample have a small number.

That is the reason of why I became interested in GADatedClock... but I can't find more information about if it is applicable to this situation, or how it works except for your posts, and the succinct description of the script.

¿Should I abandon the analysis of direct sequences?

Thank you for your patience

Andrés