Recent Approaches for Detecting Local Selection
Event details
| Date | 02.04.2014 |
| Hour | 15:00 |
| Speaker | Prof. Mark Beaumont, University of Bristol (UK) |
| Location | |
| Category | Conferences - Seminars |
BIOENGINEERING SEMINAR
Abstract:
In the past 5 years there has been an increased use of methods for detecting the putative effects of natural selection that lead to adaptive differences between populations. At the same time there has been a greatly increased appreciation of the difficulties inherent in such methods, particularly those that lead to false positives. This talk will give a brief review of the area, starting with Lewontin and Krakauer's original proposals. Most methods are based on detecting outliers under a neutral model of differentiation. I will describe a recent approach (Vitalis, Gautier, Dawson & Beaumont, Genetics, 2014) in which the parameterisation is in terms of Wright's stationary distribution for alleles under selection in an infinite island model. The method appears to have some advantages in terms of ROC characteristics, and reduced sensitivity to the effects of population covariance in allele frequency.
Bio:
Education:
- BSc Zoology (Manchester)
- PhD Genetics (Nottingham)
Honours:
- Wellcome Mathematical Biology Fellowship (1992-1995)
- NERC Advanced Fellowship (2003-2008)
Research Interests:
I am a biologist by background, and I am interested in general problems of statistical inference in population genetics, evolutionary biology, and conservation genetics. Most of my work has involved Monte Carlo statistical methods. Particular areas of application that interest me include: detecting evidence of selection in the genome; modelling demographic history of populations; inference in structured populations; modelling temporally sampled genetic data; inference in agent-based models.
Abstract:
In the past 5 years there has been an increased use of methods for detecting the putative effects of natural selection that lead to adaptive differences between populations. At the same time there has been a greatly increased appreciation of the difficulties inherent in such methods, particularly those that lead to false positives. This talk will give a brief review of the area, starting with Lewontin and Krakauer's original proposals. Most methods are based on detecting outliers under a neutral model of differentiation. I will describe a recent approach (Vitalis, Gautier, Dawson & Beaumont, Genetics, 2014) in which the parameterisation is in terms of Wright's stationary distribution for alleles under selection in an infinite island model. The method appears to have some advantages in terms of ROC characteristics, and reduced sensitivity to the effects of population covariance in allele frequency.
Bio:
Education:
- BSc Zoology (Manchester)
- PhD Genetics (Nottingham)
Honours:
- Wellcome Mathematical Biology Fellowship (1992-1995)
- NERC Advanced Fellowship (2003-2008)
Research Interests:
I am a biologist by background, and I am interested in general problems of statistical inference in population genetics, evolutionary biology, and conservation genetics. Most of my work has involved Monte Carlo statistical methods. Particular areas of application that interest me include: detecting evidence of selection in the genome; modelling demographic history of populations; inference in structured populations; modelling temporally sampled genetic data; inference in agent-based models.
Practical information
- Informed public
- Free