Tools to Support Phylogenetic Analysis

Introduction

Over the coming year FIG and its friends plan on developing and releasing a number of tools to support phylogenetic analysis. At this point, we are making available a tool for inserting taxa into a tree, assuming that one has an alignment of SSU rRNA and a tree that includes some subset of the taxa in the alignment. At one time this technology was used to extend the tree distributed by thye Ribosomal Database Project. We have revived it and make it available now as a first step in supporting the development of a large SSU-based phylogenetic tree. By itself, it is not adequate for many tasks. A few key tools are needed to complement the set we are making available. We plan on making these additional tools available over the coming year.

Extending an Existine Tree by Insertion of One Sequence at a Time

Assuming that you have

a new alignment (call it ssu_alignment.fasta),
a table giving a correspondence between IDs and organisms (call it ssu.names), and
an old tree (call it old.ssu.tree).

you should follow these steps to extend the tree:

First, you need to verify that all of the ids in the tree are still in the alignment. To do this, run compare_tree_and_alignment ssu_alignment.fasta old.ssu.tree tmp.tree.only tmp.ali.only tmp.both

If tmp.tree.only is not empty, run mv old.ssu.tree old.ssu.tree.BAK subtree_of old.ssu.tree.Bak < tmp.both > old.ssu.tree

If you are inserting into a tree that may contain fragments, you should probably consider removing short sequences and then insert sequences in descending order of length (i.e., nonambiguous characters). count_bases < ssu_alignment.fasta | sort -n -r +1 > nonambiguous
gives counts for sequences in the alignment. initial_set tmp.both nonambiguous > initial.ids subtree_of old.ssu.tree < initial.ids > initial.tree mv initial.tree old.ssu.tree to_insert nonambiguous initial.ids > tmp.ali.only
is how we recommend handling this.
Now you need to get weights and rates for each column of the alignment. To do this, run make_rates ssu_alignment.fasta old.ssu.tree > weights_and_rates
Now you are ready to do insertions. To accomplish this, run insert_all ssu_alignment.fasta old.ssu.tree weights_and_rates < tmp.ali.only > new.ssu.tree
As the insertion runs, it updates the "old.ssu.tree". This means that if the run gets terminated, take off the initial section of tmp.ali.only, and just restart it.

You can display your tree (very crudely) using display_tree new.ssu.tree ssu.names
Note that the tree is unrooted. We supply a command for rooting it, but you do need to understand exactly where you wish to place the root. To root it, use root_at Node1 Node2 FractionBetween < UntootedTree > RootedTree
where the the nodes are specified as either

tip id, or
three ids separated by commas (which gives a unique point in the tree).

You can extract a representative tree by using representative_tree big.tree N
where big.tree is the file containing a newick tree and N is the number of nodes desired in the representative tree.