Comments on: You know, people do use neighbour joining!

By: Mailund on the Internet » It is not all bad news

Mailund on the Internet » It is not all bad news — Thu, 19 Jun 2008 17:17:46 +0000

[...] it is on neighbour-joining, we weren’t that optimistic. We’ve had problems publishing on this before, but this time it was very well [...]

By: Thomas Mailund

Thomas Mailund — Thu, 27 Mar 2008 08:42:45 +0000

You always need to treat inferred data with a bit of caution, of course. Regardless of how you infer. For a ML tree you would have a sound statistical framework for this, while for a NJ tree you would rely on bootstrap or such.

In any case, you always want to go for the most reliable method feasible to use, and my point is just that for large data sets, that is not ML trees. You run into a tradeoff between trusting the method or relying on more data, and usually you want to use as much data as you can get your hands on as well… Ah, I don’t know… it is a tricky question ;-)

By: Bob O'H

Bob O'H — Thu, 27 Mar 2008 04:18:53 +0000

Hear, hear!

Of course this does raise questions about whether the trees are any good, but I guess you jut have to bear that in mind when interpreting them.