Limits...
TreeKO: a duplication-aware algorithm for the comparison of phylogenetic trees.

Marcet-Houben M, Gabaldón T - Nucleic Acids Res. (2011)

Bottom Line: To overcome this, we have developed treeKO, an algorithm that enables the comparison of tree topologies, even in the presence of duplication and loss events.To do so treeKO recursively splits gene trees into pruned trees containing only orthologs to subsequently compute a distance based on the combined analyses of all pruned tree comparisons.In addition treeKO, implements the possibility of computing phylome support values, and reconciliation-based measures such as the number of inferred duplication and loss events.

View Article: PubMed Central - PubMed

Affiliation: Centre for Genomic Regulation, UPF, Doctor Aiguader 88, 08003 Barcelona, Spain.

ABSTRACT
Comparisons of tree topologies provide relevant information in evolutionary studies. Most existing methods share the drawback of requiring a complete and exact mapping of terminal nodes between the compared trees. This severely limits the scope of genome-wide analyses, since trees containing duplications are pruned arbitrarily or discarded. To overcome this, we have developed treeKO, an algorithm that enables the comparison of tree topologies, even in the presence of duplication and loss events. To do so treeKO recursively splits gene trees into pruned trees containing only orthologs to subsequently compute a distance based on the combined analyses of all pruned tree comparisons. In addition treeKO, implements the possibility of computing phylome support values, and reconciliation-based measures such as the number of inferred duplication and loss events.

Show MeSH
Distribution of distances between trees in P12a phylome and three alternative species trees. The upper left part of the figure shows the three topologies used. The first one is the T12a tree while the other two represent changes in this topology. Alternative topology 1 represents a change in a poorly supported node while Alternative topology 2 represents a well supported node. The two upper right graphs plot each distribution of distances of the alternative topologies against the reference T12a topology. The lower panel represents the frequency graph for the three distance distributions.
© Copyright Policy - creative-commons
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC3105381&req=5

Figure 2: Distribution of distances between trees in P12a phylome and three alternative species trees. The upper left part of the figure shows the three topologies used. The first one is the T12a tree while the other two represent changes in this topology. Alternative topology 1 represents a change in a poorly supported node while Alternative topology 2 represents a well supported node. The two upper right graphs plot each distribution of distances of the alternative topologies against the reference T12a topology. The lower panel represents the frequency graph for the three distance distributions.

Mentions: Although phylomes have successfully been used to determine which nodes in the fungal species tree are most congruent at genomic-scales (22), there is as yet no RF-based measure of the levels of similarity between a given species tree and a complete phylome. Here, we address the question of whether the distributions of speciation distances to a given phylome can be used to decide among alternative evolutionary scenarios. For this we used treeKO to compute the speciation distances of the yeast phylome [P12a dataset described in (22) and available at PhylomeDB (32)] against different alternative species trees. Gene trees were rooted with midpoint so that no assumptions on the species tree topology were made a priori. Alternative topologies were derived from a reference species tree by swapping pairs of neighboring branches (i.e. interchanging the positions of Saccharomyces paradoxus and Saccharomyces mikatae or the positions of C. glabrata and S. castellii as shown in Figure 2 and Supplementary Figure S1). Resulting distance distributions were compared with a t-test.Figure 2.


TreeKO: a duplication-aware algorithm for the comparison of phylogenetic trees.

Marcet-Houben M, Gabaldón T - Nucleic Acids Res. (2011)

Distribution of distances between trees in P12a phylome and three alternative species trees. The upper left part of the figure shows the three topologies used. The first one is the T12a tree while the other two represent changes in this topology. Alternative topology 1 represents a change in a poorly supported node while Alternative topology 2 represents a well supported node. The two upper right graphs plot each distribution of distances of the alternative topologies against the reference T12a topology. The lower panel represents the frequency graph for the three distance distributions.
© Copyright Policy - creative-commons
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC3105381&req=5

Figure 2: Distribution of distances between trees in P12a phylome and three alternative species trees. The upper left part of the figure shows the three topologies used. The first one is the T12a tree while the other two represent changes in this topology. Alternative topology 1 represents a change in a poorly supported node while Alternative topology 2 represents a well supported node. The two upper right graphs plot each distribution of distances of the alternative topologies against the reference T12a topology. The lower panel represents the frequency graph for the three distance distributions.
Mentions: Although phylomes have successfully been used to determine which nodes in the fungal species tree are most congruent at genomic-scales (22), there is as yet no RF-based measure of the levels of similarity between a given species tree and a complete phylome. Here, we address the question of whether the distributions of speciation distances to a given phylome can be used to decide among alternative evolutionary scenarios. For this we used treeKO to compute the speciation distances of the yeast phylome [P12a dataset described in (22) and available at PhylomeDB (32)] against different alternative species trees. Gene trees were rooted with midpoint so that no assumptions on the species tree topology were made a priori. Alternative topologies were derived from a reference species tree by swapping pairs of neighboring branches (i.e. interchanging the positions of Saccharomyces paradoxus and Saccharomyces mikatae or the positions of C. glabrata and S. castellii as shown in Figure 2 and Supplementary Figure S1). Resulting distance distributions were compared with a t-test.Figure 2.

Bottom Line: To overcome this, we have developed treeKO, an algorithm that enables the comparison of tree topologies, even in the presence of duplication and loss events.To do so treeKO recursively splits gene trees into pruned trees containing only orthologs to subsequently compute a distance based on the combined analyses of all pruned tree comparisons.In addition treeKO, implements the possibility of computing phylome support values, and reconciliation-based measures such as the number of inferred duplication and loss events.

View Article: PubMed Central - PubMed

Affiliation: Centre for Genomic Regulation, UPF, Doctor Aiguader 88, 08003 Barcelona, Spain.

ABSTRACT
Comparisons of tree topologies provide relevant information in evolutionary studies. Most existing methods share the drawback of requiring a complete and exact mapping of terminal nodes between the compared trees. This severely limits the scope of genome-wide analyses, since trees containing duplications are pruned arbitrarily or discarded. To overcome this, we have developed treeKO, an algorithm that enables the comparison of tree topologies, even in the presence of duplication and loss events. To do so treeKO recursively splits gene trees into pruned trees containing only orthologs to subsequently compute a distance based on the combined analyses of all pruned tree comparisons. In addition treeKO, implements the possibility of computing phylome support values, and reconciliation-based measures such as the number of inferred duplication and loss events.

Show MeSH