Thursday, March 27, 2008

Make Your Own Phylogenetic Tree

 
Sandra Porter at Discovering Biology in a Digital World has prepared a short video presentation on how to make your very own phylogenetic tree from DNA sequences [A beginner's guide to making a phylogenetic tree].


3 comments :

  1. Its too bad she does not know the difference between phylogenetics and Cladistic. She uses NJ, which is a cladistic clustering method based on overall similarity. These methods do not reconstruct the evolutionary history of anything, but only cluster sequences based on percent similarity, effectively ignoring synapomorphies etc. that are crucial in phylogenetics analyses such as maximum parsimony.

    ReplyDelete
  2. I made so many mistakes in my precious post that I think I need to repost it:
    Its too bad she does not know the difference between phenetics and Cladistic. She uses NJ, which is a phenetic clustering method based on overall similarity. These methods do not reconstruct the evolutionary history of anything, but only cluster sequences based on percent similarity, effectively ignoring synapomorphies etc. that are crucial in cladistic (phylogenetic) analyses such as maximum parsimony.

    ReplyDelete
  3. Neighbor Joining (NJ) works quite well for most data and difference methods are easy for students to understand.

    The are many more complicated methods for constructing phylogenetic trees and some of them have good theoretical support. There are reasons to believe that some of these methods work better than difference methods when the data is ideal.

    Unfortunately, the data (sequences) are not as reliable as most people believe and correct multiple sequence alignments are quite a challenge.

    If you are using a program like ClustlW to align sequences then you are introducing far larger error bars into your data than the difference between tree building programs.

    It's rather silly to be debating the superiority of NJ vs. maximum parsimony or maximum likelihood when the alignments are so much more likely to be the major source of error.

    ReplyDelete