Phylip Distance Matrix Question

Question

[Deena]

Hello, I am using Phylipv3.69 under Linux. I have two questions regarding generating the distance matrix using protdist:

1. Is there a way to output the distance matrix as a lower triangular similarity matrix, instead of a square matrix? I don't find this option within the menu for protdist.

2. I generated a neighbor-joining majority-rule consensus tree using seqboot->protdist->neighbor->consense. Now that the consense has picked the final tree from 100 bootstrapped tree, is there a way to go back and determine the precise distance matrix that was used to generate that tree? Or do the branch lengths on the tree suffice as the distance matrix in that case? Apologies if this seems to be a naive question, as I am not very familiar with phylogenetic analyses. The idea is that I generated two trees and I wanted to get the distance matrices of the the trees to compare them.

Thank you in advance for all your help and advice!

Answer 1

[David W]

Hi Deena, I don't think you're going to like my answers...

1. I don't think so, dnadist has an option for a triangular matrix but can't see one for protdist

2. I don't think so again. The pipeline you used made 100 (pseudo)samples of your original aliment, then calculated the distances for each. So there isn't just one distance matrix underlying it. Consense puts the set of 100 trees together into one, but that tree doesn't actually have branch lengths. (I think if you had one tree, it was estimated with nj and it didn't have any polytomies you could go backwards from branch lengths to distances, but that's off to top of my head and shouldn't be taken too seriously ;-).

BTW, although I actually quite like phylip, it's not that widely used anymore. There are quite a lot of discussion on "the best" distance-based phylogenetic packages on this site, you might find a program you like more

[Michael Schubert]

If there are only pairwise distance values in the matrix, dist(A,B)=dist(B,A) which means that the matrix is symmetrical. To get the lower diagonal matrix, just ignore values above the main diagonal.

Answer 2

[Deena]

Hi David, Thank you for your responses. I guess I will just have to be content with the branch length values in my final tree, and forget about the distance matrix! Deena