Biology – Quantitative Biology – Quantitative Methods
Scientific paper
2010-04-26
Biology
Quantitative Biology
Quantitative Methods
11 pages, 2 figures
Scientific paper
This paper combines concepts from phylogenetics with Gaussian process regression, in order to allow evolutionary inference for function-valued traits that are correlated through phylogeny. By function-valued traits, we mean data objects which are indexed by one or more `spatial' co-ordinates, such as organism age or buffer pH level. Examples of function-valued traits include organism mass with age or fitness vs pH curves. We provide a nonparametric Bayesian model for such data by using Gaussian processes. The model may be used to infer ancestral function-valued traits, compare rates of evolution across a phylogeny, or to identify the most likely phylogenies consistent with observed data. It contrasts with methods which reduce data to summary statistics or a multivariate vector (without spatial co-ordinates) and allows us to make inferential statements about ancestral traits themselves. We illustrate the use of the method on real data we generated by experiment.
Jones Nick S.
Moriarty John
No associations
LandOfFree
Evolutionary Inference for Function-valued Traits: Gaussian Process Regression on Phylogenies does not yet have a rating. At this time, there are no reviews or comments for this scientific paper.
If you have personal experience with Evolutionary Inference for Function-valued Traits: Gaussian Process Regression on Phylogenies, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Evolutionary Inference for Function-valued Traits: Gaussian Process Regression on Phylogenies will most certainly appreciate the feedback.
Profile ID: LFWR-SCP-O-236581