hit counter joomla

If you’ve ever looked at an evolutionary tree, contemplated phylogeny, cladistics, or the like, you’re probably aware that Joe Felsenstein is one of the leaders of the pack.

And you will certainly enjoy, this interview that Blind Scientist has posted.

I wouldn’t advise reading the interview to learn about doing phylogenetics, but you will learn a bit about the social anthropology of the field. Felsenstein does a wonderful job of supplying historical context to phylogenetic arguments and filling in the missing details.

A phylogenetic tree is a kind of model or hypothesis that’s put forth to describe the evolutionary history of certain species. In the days before high-throughput DNA sequencing, people would propose trees and argue about them, but no one could really know which tree was true. The proponents of parsimony were adamant that evolution could only proceed by the most mathematically-pleasing process. In other words, they argued that the tree, or evolutionary path with the fewest number of changes must, by necessity, be the best. (The most parsimonious tree is one with the fewest changes.)

As a biologist, of course, I never understood that sort of logic. And, the experiments confirmed that biological systems didn’t grasp it either.

To me, it made much more sense to use models that recognized the existence of chemistry and agree that some changes between nucleotides happen more often than others.

Since those days, experimental data have demonstrated that the maximum likelihood algorithm in Felsentein’s Phylip package does produce better trees. Maximum-likelihood trees mimic reality the best, with the neighbor-joining algorithm doing pretty well, too, and parsimony, last.

JF also shares his opinions on open source software and corrects some misconceptions. Phylip, for example, like other famous bioinformatics programs developed at the UW, is not open source and must be licensed from the University for commercial use.

Whether you agree with his ideas or not, JF’s reasons for preferring that route are worth a read.


  1. #1 Ying-tsu Loh
    May 10, 2007

    Hi Sandy,
    I’m glad you brought up this topic. I’ve been trying to do some phylogenetic trees using some sequences such as cytochrome c, lysosome or myoglobin. I haven’t been able to find a good source for basic phylogenetics. For example, some papers say it’s best to chop off areas that don’t have good matches in the multiple sequence alignment (gaps, etc) Of course, once that is done, the tree looks very different (I’ve been using ClustalW then going into Jalview to do my nj trees). Do you have any suggestions for a good source for basic phylogenetics (Bioinformatics for Dummies and the Baxevanis books were not very clear I thought) or should I forget about Jalview for trees and go directly into PHYLIP?

  2. #2 Sandra Porter
    May 10, 2007

    I tried to give a more detailed answer and my comment disappeared! I’ll put an answer in a separate post.

  3. #3 Paulo Nuin
    June 25, 2007

    Hi Sandra

    Thanks a lot for commenting on the interview. I finished what I call the inagural cycle of the SciView interviews with a couple of other scientists: Brian Golding and Jeremy Squire. Great advices for the young scientist. The direct links are below:



    I hope to keep the project alive and interview many others in the near future.

    Thanks for the interest and support.

    Paulo Nuin

New comments have been temporarily disabled. Please check back soon.