Department of Statistics Seminar
North Carolina State University

presents

Dr. Asger Hobolth

North Carolina State University

"Statistical inference in evolutionary models of DNA sequences"

ABSTRACT

Large-scale DNA sequencing efforts, such as the Human, Mouse and Chicken Genome Projects, are producing an immense amount of data. The need to understand the underlying evolutionary processes, that shape the data, is becoming ever more pressing. Comparison of sequence data from contemporary species are increasingly being based on probabilistic models. One such model is the continuous time Markov model. A continuous time Markov model is characterized by a rate matrix that describes the time between substitutions and the type of substitution that may occur. Several problems arise when continuous time Markov models are used for describing the evolution of DNA sequence data. Firstly, we only observe data from contemporary species and information on substitution events on the phylogenetic tree, that relates the species, are missing. Secondly, the evolution of each letter in the DNA sequence is neighbour dependent, leading to huge rate matrices that are difficult to handle. In this talk I will review and discuss recent advances in statistical inference for continuous time Markov models with applications to biological sequence data.

Friday, September, 2, 2005

3:35 - 4:35 pm

206 Cox Hall

Refreshments will be served on the second floor of Dabney Hall (left of Room 222) at 3:00 pm.