IC Monday Seminars : Bridging the Gap between Individual and Population Views of Intraspecies Phylogenetics
Event details
| Date | 30.04.2012 |
| Hour | 16:15 |
| Speaker | Prof. Russell Schwartz (CMU) - Hosted by Prof. B. Moret |
| Location |
INM 202
|
| Category | Conferences - Seminars |
Abstract
Modern high-throughput sequencing technologies have enabled a series of major efforts to catalog intraspecies genetic variations, dramatically increasing the amount of variation data available for humans and other well studied species. In principle, such data should allow us to characterize the process of molecular evolution within single species with unprecedented precision. In practice, however, there is a large and growing gap between the amount of data we have available and our ability to use it productively.
This talk will describe our effort to address this gap by combining elements of two major approaches to understanding genetic variation data: a phylogenetic view, which seeks to identify evolutionary trees linking genotyped individuals, and a population view, which seeks to categorize individuals or pieces of their chromosomes into population subgroups defined by common recent ancestry. Each view brings to the problem its own specialized models and algorithms in order to characterize the history of a species in different ways. By combining elements of both methods, we develop an approach for the problem of automatically inferring ``population histories’’ that seek to characterize both the robustly supported population substructure of a species and the sequence of divergence and admixture events by which that substructure developed. In this talk, we will first discuss the motivation behind this work and the need for new methods for the problem. We will then explore computational issues in tackling the problem by combining discrete algorithmic approaches from the field of phylogenetics with machine learning and statistical sampling approaches adapted from statistical genetics. Finally, we will examine how such methods work in practice on both simulated and real genetic variation data and what they can tell us about the likely population history of the human species.
Biography
Russell Schwartz received his Ph.D. in Computer Science at the Massachusetts Institute of Technology in 2000. He then worked at Celera Genomics on genetic variation analysis before joining Carnegie Mellon University in 2002, where he is currently a Professor in the Department of Biological Sciences and the Lane Center for Computational Biology. His lab studies various problems in computational biology, including phylogenetics and population genetics, cancer biology, and simulation of complex biophysical systems.
Modern high-throughput sequencing technologies have enabled a series of major efforts to catalog intraspecies genetic variations, dramatically increasing the amount of variation data available for humans and other well studied species. In principle, such data should allow us to characterize the process of molecular evolution within single species with unprecedented precision. In practice, however, there is a large and growing gap between the amount of data we have available and our ability to use it productively.
This talk will describe our effort to address this gap by combining elements of two major approaches to understanding genetic variation data: a phylogenetic view, which seeks to identify evolutionary trees linking genotyped individuals, and a population view, which seeks to categorize individuals or pieces of their chromosomes into population subgroups defined by common recent ancestry. Each view brings to the problem its own specialized models and algorithms in order to characterize the history of a species in different ways. By combining elements of both methods, we develop an approach for the problem of automatically inferring ``population histories’’ that seek to characterize both the robustly supported population substructure of a species and the sequence of divergence and admixture events by which that substructure developed. In this talk, we will first discuss the motivation behind this work and the need for new methods for the problem. We will then explore computational issues in tackling the problem by combining discrete algorithmic approaches from the field of phylogenetics with machine learning and statistical sampling approaches adapted from statistical genetics. Finally, we will examine how such methods work in practice on both simulated and real genetic variation data and what they can tell us about the likely population history of the human species.
Biography
Russell Schwartz received his Ph.D. in Computer Science at the Massachusetts Institute of Technology in 2000. He then worked at Celera Genomics on genetic variation analysis before joining Carnegie Mellon University in 2002, where he is currently a Professor in the Department of Biological Sciences and the Lane Center for Computational Biology. His lab studies various problems in computational biology, including phylogenetics and population genetics, cancer biology, and simulation of complex biophysical systems.
Practical information
- Informed public
- Free
Organizer
- Simone Muller
Contact
- Simone Muller