Phylogenetic Tree Visualization: Challenges and Solutions

1. Phylogenetic Tree Visualization: Challenges and Solutions Md Liakat Aliand Sung-Hyuk Cha May 1, 2015

2. Contents • Introduction • Challenges in visualizing phylogenetic trees • Tree presentation and its combinatorics • Survey on crossings minimization problem • Solution for crossings minimization

3. Introduction • Phylogenetics relies on exploring the ways of visual inspection, structural comparison and manipulation of phylogenies using phylogenetic trees. • Different conflicting trees from same dataset can produce by different algorithms, techniques, and different measures. So to compare them is a very hard problem • How to choose most pleasing or meaningful tree from large number of isomorphic trees.

4. Challenges in visualizing Phylogenetic Trees • Main challenges are: • Layout- important for designing algorithms, because complexity differs with different tree styles. • Labeling and annotation- should be readable, related to graphics object, and should not obscure any pertinent information. • Navigation- • Some techniques and algorithms: focus+context, ATV, TreeJuztaposer, and TreeWiz. • Not suitable for large tree, growing some areas while shrinking others, creates occlusion and expensive.

5. Challenges in visualizing Phylogenetic Trees • Main challenges are: • Tree comparison • Some techniques: consensus tree, nearest neighbor interchange, agreement sub-tree, triplets. • Some tools: COMPONENT, FastDNAml, TreeSet, PHYLIP, Mesquite, T-Rex • Do not offer good distribution of similarity score, highly sensitive to topology of the tree, not time efficient, bad resolution. • Manipulation and editing • Output format of programs are not suitable for publication and need to edit manually or editing software.

6. Tree presentation and its combinatorics TABLE I A sample distance matrix of five taxa Equations used to measure the distance between subsets in single linkage and complete linkage method

7. Tree presentation and its combinatorics Default dendrogram T1 of Table 1 using single linkage method. Default dendrogram T2 of Table 1 using complete linkage method.

8. Tree presentation and its combinatorics • TABLE II (a) Dendrogram representation for T1 • TABLE II (b) Dendrogram representation for T2

9. Tree presentation and its combinatorics Permuted dendrograms for Table II (a) without crossings Permuted dendrograms for Table II (a) with crossings

10. Tree presentation and its combinatorics Displaying two dendrograms: T1 on top and T2 below

11. Previous Studies in Crossings Minimization Algorithm 1 to minimize the number of crossings and algorithm 2 is to count number of crossings between two nodes proposed by Dwyer and Schreiber

12. Previous Studies in Crossings Minimization

13. Proposed Solution for crossings minimization • For a given dendrogram and given leaf node order, we have to consider two important constraints. • All leaf nodes must be same bottom level and order should be specified. • All internal nodes heights must correspond to the user’s defined distance between two clusters either exactly or at least order-wise.

14. Proposed Solution for crossings minimization • Steps: • Find which two clusters have maximum number of crossings. • Merge them- left or right (Choose that minimizes crossings) • If there are no crossings or number of crossings tie, it will simply merge the closest pair of clusters. • The processes repeated recursively

15. Proposed Solution for crossings minimization Example of Arbitrary order such as <C, E, A, D, B> < A, B, C, D, E>

16. Proposed Solution for crossings minimization Example of Arbitrary order such as <C, A, D, E, B>

17. Thank You ?