7. Phylogenetic prediction

Find the evolutionary tree or trees that best accounts for the observed variation (mismatches) in the multiple alignment of a group of sequences.


[few sequences, low variation]
  1. Character based: Maximum parsimony, compatibility
  2. Distance based: Neighbour-joining, UPGMA
  3. Probabilistic approaches: Maximum likelihood
[many sequences, high variation]

For n species, there are 1 x 3 x 5 x ... x (2n - 3) rooted trees

Bootstrap analysis of trees: Reliability of a predicted phylogenetic tree.

  1. Resampling data by random selection of columns in the msa to create a new one
  2. Compute the phylogenetic tree with the same method
  3. Significant branches should frequently appear in the resampled trees