Your browser doesn't support javascript.
Spectral partitioning of phylogenetic data sets based on compatibility.
Syst Biol ; 56(4): 623-32, 2007 Aug.
Artigo em Inglês | MEDLINE | ID: mdl-17654366
ABSTRACT
We describe two new methods to partition phylogenetic data sets of discrete characters based on pairwise compatibility. The partitioning methods make no assumptions regarding the phylogeny, model of evolution, or characteristics of the data. The methods first build a compatibility graph, in which each node represents a character in the data set. Edges in the compatibility graph may represent strict compatibility of characters or they may be weighted based on a fractional compatibility scoring procedure that measures how close the characters are to being compatible. Given the desired number of partitions, the partitioning methods then seek to cluster the characters with the highest average pairwise compatibility, so that characters in each cluster are more compatible with each other than they are with characters in the other cluster(s). Partitioning according to these criteria is computationally intractable (NP-hard); however, spectral methods can quickly provide high-quality solutions. We demonstrate that the spectral partitioning effectively identifies characters with different evolutionary histories in simulated data sets, and it is better at highlighting phylogenetic conflict within empirical data sets than previously used partitioning methods.
Assuntos

Similares

MEDLINE

...
LILACS

LIS

Coleções: Bases de dados internacionais Base de dados: MEDLINE Assunto principal: Filogenia / Simulação por Computador / Software Idioma: Inglês Revista: Syst Biol Assunto da revista: Biologia Ano de publicação: 2007 Tipo de documento: Artigo País de afiliação: Estados Unidos