General phylogenetic tree package

Question

[shai2go]

I am looking for a package that will allow me to use phylogenetic tree creation algorithms, with more general data than fasta format for genetic data.

I found lots of packages but they require .fasta format and work for amino acids or nucleotides.

I need to be able to define the traits of the data, and the way of distance calculation.

I prefer something that will work in python and if not python or r.

Thanks

[Medhat]

what kind of format you have?

[shai2go]

i am trying to construct phylogenetic tree for texts, so the format is somewhat more complex, a matrix of n*m, every matrix(i,j) is a word. i have a component that will calculate the distance beetween different entries of the matrix.

i can reduce the words to vector of values and can calculate distance beetween them. i need to be able to activate the phylogenetic algorithm with my own distance calculation.

@medhat

[Medhat]

is this do the job?

http://pycogent.org/cookbook/building_phylogenies.html

http://etetoolkit.org/docs/latest/tutorial/index.html

http://biopython.org/wiki/Phylo

[shai2go]

@medhat

all the above are good for 1.inferring trees for genetic information 2.plotting trees that are allready made.

i need to infer trees for data, that is not genetic, hence, i need to be able to create the distance matrix for it. i can't find any way to do so with what you posted...

Answer 1

[Jean-Karim Heriche]

It seems you're looking to draw a phylogenetic tree using an arbitrary distance matrix as input. A number of phylogeny software can do this but they normally use generic hierarchical clustering algorithms such as implemented in the agnes() function of the R package cluster. If you really want to use a phylogeny software, you can try the R packages ape or phangorn. They have various neighbour-joining functions taking a distance matrix as input.

[shai2go]

@Jean-karim - thanks! ape is what i wanted, nj and mvr do the trick, but a little issue, i need to construct a tree assuming that let's say, if i get 1,2,3 there is a possibility of 1 being the father of 2,3 (that are brothers). nj and mvr give me a tree where all the samples are leaves, and none of them are the father of one another,

is there a way not to assume that?

thanks.