| Statistical Programs |
| College of Agriculture | University of Idaho |
| Seminar Announcement |
| "Applied Statistics in Agriculture" |
|
Bootstrap Estimation and Inference on an Index of Phylogenetic Correlation
Presented By |
| Dr. William J. Price |
|
Statistical Programs College of Agriculture University of Idaho |
|
Tuesday, November 17 3:30 P. M. Ag. Science 62 |
| A common objective of bioinformatic analyses is to assess the similarity of species or genotypic variations. Such measures provide a means to evaluate evolutionary models and history as well as having potential application to ecological systems such as host preference selection. Phylogenetic correlation (1) is one index of similarity typically employed. Given a phylogeny, the correlation, , measures the deviation or adjustment relative to a dependent Brownian evolutionary model. Values for are found through maximum likelihood estimation of a generalized linear model assuming a variance-covariance structure, where off diagonal elements are scaled by . A value of = 1.0 is indicative of the Brownian model, while = 0.0 indicates an independent random process. Statistical inference on has traditionally been assessed using a likelihood ratio test comparing the estimated value to the theoretical null values, =1.0 and =0.0. These tests, however, rely on the assumption of a Normal likelihood within the phylogeny. In addition, statistical comparison of estimated values has not been addressed. An alternative procedure is proposed here which relies instead on the resampling methodology of the bootstrap. Here, is estimated with an underlying bootstrap distribution which provides both point estimation and associated confidence limits. The method will be demonstrated using phylogenetic and metabolomic data related to the host specificity of Ceutorhynchus cardariae Korotyaev on Brassicaceae species. |
| Return to Seminar Listing. |