TY - GEN
T1 - Reducing distortion in phylogenetic networks
AU - Huson, Daniel H.
AU - Steel, Mike A.
AU - Whitfield, Jim
PY - 2006/1/1
Y1 - 2006/1/1
N2 - When multiple genes axe used in a phylogenetic study, the result is often a collection of incompatible trees. Phylogenetic networks and super-networks can be employed to analyze and visualize the incompatible signals in such a data set. In many situations, it is important to have control over the amount of imcompatibility that is represented in a phylogenetic network, for example reducing noise by removing splits that do not recur among the source trees. Current algorithms for computing hybridization networks from trees are based on a combinatorial analysis of the arising set of splits, and are thus sensitive to false positive splits. Here, a filter is desirable that can identify and remove splits that are not compatible with a hybridization scenario. To address these issues, the concept of the distortion of a tree relative to a split is defined as a measure of how much the tree needs to be modified in order to accommodate the split, and some of its properties are investigated. We demonstrate the usefulness of the approach by recovering a plausible hybridization scenario for buttercups from a pair of gene trees that cannot be obtained by existing methods. In a second example, a set of seven gene trees from microgastrine braconid wasps is investigated using filtered networks. A user-friendly implementation of the method is provided as a plug-in for the program SplitsTree4.
AB - When multiple genes axe used in a phylogenetic study, the result is often a collection of incompatible trees. Phylogenetic networks and super-networks can be employed to analyze and visualize the incompatible signals in such a data set. In many situations, it is important to have control over the amount of imcompatibility that is represented in a phylogenetic network, for example reducing noise by removing splits that do not recur among the source trees. Current algorithms for computing hybridization networks from trees are based on a combinatorial analysis of the arising set of splits, and are thus sensitive to false positive splits. Here, a filter is desirable that can identify and remove splits that are not compatible with a hybridization scenario. To address these issues, the concept of the distortion of a tree relative to a split is defined as a measure of how much the tree needs to be modified in order to accommodate the split, and some of its properties are investigated. We demonstrate the usefulness of the approach by recovering a plausible hybridization scenario for buttercups from a pair of gene trees that cannot be obtained by existing methods. In a second example, a set of seven gene trees from microgastrine braconid wasps is investigated using filtered networks. A user-friendly implementation of the method is provided as a plug-in for the program SplitsTree4.
UR - http://www.scopus.com/inward/record.url?scp=33750265162&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=33750265162&partnerID=8YFLogxK
U2 - 10.1007/11851561_14
DO - 10.1007/11851561_14
M3 - Conference contribution
AN - SCOPUS:33750265162
SN - 3540395830
SN - 9783540395836
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 150
EP - 161
BT - Algorithms in Bioinformatics - 6th International Workshop, WABI 2006, Proceedings
PB - Springer
T2 - 6th International Workshop on Algorithms in Bioinformatics, WABI 2006
Y2 - 11 September 2006 through 13 September 2006
ER -