Using phylogeny application controllers to construct phylogenetic trees from alignmentsΒΆ

Section author: Daniel McDonald

This document provides a few use case examples of how to use the phylogeny application controllers available in PyCogent. Each phylogeny application controller provides the support method build_tree_from_alignment. This method takes as input an Alignment object, a SequenceColleciton object or a dict mapping sequence IDs to sequences. The MolType must also be specified. Optionally, you can indicate if you would like the “best_tree$”, as well as any additional application parameters. These methods return a PhyloNode object.

To start, lets import all of our build_tree_from_alignment methods and our MolType:

>>> from cogent.core.moltype import DNA
>>> from import build_tree_from_alignment as clearcut_build_tree
>>> from import build_tree_from_alignment as clustalw_build_tree
>>> from import build_tree_from_alignment as fasttree_build_tree
>>> from import build_tree_from_alignment as muscle_build_tree
>>> from import build_tree_from_alignment as raxml_build_tree

Next, we’ll load up a test set of sequences and construct an Alignment:

>>> from cogent import LoadSeqs
>>> from import align_unaligned_seqs
>>> unaligned = LoadSeqs(filename='data/test2.fasta', aligned=False)
>>> aln = align_unaligned_seqs(unaligned, DNA)

Now, let’s construct some trees with default parameters!


We are explicitly seeding Clearcut and RAxML to ensure reproducible results, and FastTree’s output depends slightly on which version of FastTree is installed

>>> clearcut_tree = clearcut_build_tree(aln, DNA, params={'-s':42})
>>> clustalw_tree = clustalw_build_tree(aln, DNA)
>>> fasttree_tree = fasttree_build_tree(aln, DNA)
>>> muscle_tree = muscle_build_tree(aln, DNA)
>>> raxml_tree = raxml_build_tree(aln, DNA, params={'-p':42})
>>> clearcut_tree
>>> clustalw_tree
>>> muscle_tree
>>> raxml_tree
>>> fasttree_tree

These methods allow the programmer to specify any of the applications parameters. Let’s look at an example where we tell Clearcut to use traditional neighbor-joining, shuffle the distance matrix, use Kimura distance correction and explicitly seed the random number generator:

>>> clearcut_params = {'-N':True,'-k':True,'-S':True,'-s':42}
>>> clearcut_tree = clearcut_build_tree(aln, DNA, params=clearcut_params)
>>> clearcut_tree