1. Computational Biology, Part D Phylogenetic Trees Ramamoorthi Ravi/Robert F. Murphy Copyright  2000, 2001. All rights reserved.

2. Phylogenetic Trees Models for process of evolution Models for process of evolution Useful Useful  finding closely related organisms that can be used as model systems  modeling HIV mutation/adaptation

3. Molecular clocks When trying to reconstruct an evolutionary tree, we wish to have a measure of “time” over which evolution has occurred When trying to reconstruct an evolutionary tree, we wish to have a measure of “time” over which evolution has occurred We can infer “time” from rates of change in sequences We can infer “time” from rates of change in sequences But not all regions of a chromosome mutate at different rates, so there is not a strict correlation between evolutionary distance and time But not all regions of a chromosome mutate at different rates, so there is not a strict correlation between evolutionary distance and time

4. Binary Trees Root: source of all other nodes Root: source of all other nodes Nodes: represents an individual protein or species Nodes: represents an individual protein or species Edges: connect nodes Edges: connect nodes Binary trees have each node connected to exactly two other nodes (children) Binary trees have each node connected to exactly two other nodes (children)

5. Building Trees Inputs = ‘sites” of variation Inputs = ‘sites” of variation  morphological characteristics (“characters”)  related genomic sequences Important to distinguish paralogues vs. orthologues Important to distinguish paralogues vs. orthologues  Need orthologues to build trees for species  Need paralogues to build trees for gene families

6. Characters as inputs for tree building species vertebrae? #legs 10 2 20 6... k1 8

7. How do this for sequence? Need to find a way to identify sites of variation in sequences Need to find a way to identify sites of variation in sequences Use a multiple sequence alignment so that each position in the sequence is a “character” Use a multiple sequence alignment so that each position in the sequence is a “character” Alternatively, find a way to measure “distance” or “dissimilarity” between sequences Alternatively, find a way to measure “distance” or “dissimilarity” between sequences

8. Tree-building methods Probabilistic Probabilistic Deterministic Deterministic

9. Probabilistic methods Assume some “random” model of evolution Assume some “random” model of evolution Find “maximum-likelihood” tree Find “maximum-likelihood” tree Only apply to “character” input Only apply to “character” input

10. Jukes-Cantor method Start with table of probabilities of any nucleotide (or amino acid) changing into each other nucleotide (or amino acid) Start with table of probabilities of any nucleotide (or amino acid) changing into each other nucleotide (or amino acid) Generate all possible trees Generate all possible trees Find that tree that gives the largest overall probability that the root sequence evolved into each of the final sequences Find that tree that gives the largest overall probability that the root sequence evolved into each of the final sequences

11. Tree-building methods Probabilistic Probabilistic  Assume some “random” model of evolution  Find “maximum-likelihood” tree  Only apply to “character” input Deterministic Deterministic  For “character” input  Parsimony  For “distance” inputs (can always be generated from “character” input)  UPGMA  Neighbor joining

12. In-class exercise Goal: Construct phylogenetic tree from a family of sequences Goal: Construct phylogenetic tree from a family of sequences “Exercise A1” on course web page (http://info.bio.cmu.edu/Courses/03310/Exe rcises/ExerciseA1.html” “Exercise A1” on course web page (http://info.bio.cmu.edu/Courses/03310/Exe rcises/ExerciseA1.html”