Coragyps

I'm not a moderator up here, but I have nominated you for sainthood and the Nobel Peace Prize on the grounds of your supression thus far. I say, "Carry on!"

Vanderzyden

Surely you'll agree that more than a simple "yes" is necessary to be persuasive about phylogenetic precision.

Vanderzyden

scigirl

Precision is always important in scientific measurements. Phylogenetics is done by using scientific measurements. ERGO precision is important in phlyogenetics!

In fact, each branch on a phylogenetic tree is associated with a number. I can't exactly remember what that number is in reference to - something about probabilities.

scigirl

Doubting Didymus

Wouldn't it be something to to with the % amount of genetic devition/similarity? I am not sure what probability would have to do with phylogeny. Could you expound?

Doubting Didymus

Sure, but you haven't told us yet what you would find convincing. I have offered to produce quotes from current university approved standard biology textbooks that state, more or less, 'phylogenetic trees agree with each other to a remarkable degree'.

I am not at home at the moment so I can't reference, but the question is would that be good enough? If you will not accept this from biology textbooks then what do you need?

lpetrich

Some methods of tree-building are probabilistic; they build up trees step-by-step, selecting which species to add at random. Not surprisingly, the final result may depend on the addition order; those probabilities are the probabilities of getting the displayed branching in different runs.

Such probabilistic methods are needed because the number of possible trees increases factorially with the number of species to be classified. For n species, the number of possible trees is 1*3*5*...*(2n-5) -- which increases VERY rapidly as n increases. And probabilistic methods are a good way of sampling the space of possible trees.

[ September 19, 2002: Message edited by: lpetrich ]</p>

Doubting Didymus

I have never heard of this method. Do you know where I can find more information?

I had always thought that trees were built straight from the molecular data.

monkenstick

vander, precision/accuracy is very much taken into consideration when phylogenetic trees are constructed, perhaps the most common method of evaluating this is a method called bootstrapping, which gives a good indication of the accuracy of nodes on phylgenetic trees

monkenstick

didymus, you could download the documentation for phylip - it explains most of the methods used to create phylogenetic trees

<a href="http://evolution.genetics.washington.edu/phylip.html" target="_blank">http://evolution.genetics.washington.edu/phylip.html</a>

Happy Wonderer

Well, maybe you are now! Repeat after me: "precision is independent of scale."
<strong>
I challenge anyone here to demonstrate even a remote comparison between the measurement of G and the "accuracy" or "precision" of phylogenetic trees.
</strong>

Sure. The "precision" of a phylogenetic tree is the probability that a tree for a group of species better represents the observed genetic data than a tree picked at random. This probability varies, of course, depending upon the number of parameters chosen and the variance observed(*). The parameters are segments of DNA sequences.


The "true" tree is a constant whose value we attempt to estimate, just like G.

(*In other words, a tree based upon one short sequence would be less accurate than a tree based on several sequences, and a tree based on a sequence that varies is more accurate than one based on a sequence that is the same for all creatures.)

If you are really interested in understanding this, <a href="http://statgen.iop.kcl.ac.uk/bgim/mle/sslike_1.html" target="_blank">here</a>is a website with a fairly good (but technical) intro to the statistical methods that seem to be commonly used in statistical genetics. (Yes, a whole new field of study for you to doubt )

HW

(Corrections and general whomping from genetic statisticans gratefully accepted. I have a very rusty background in statistics for social sciences...)