Final Report Summary - STSGA (Statistical tools for studying genetic architecture)
Biological traits are functions of a large number of genes with often complex interactions. Understanding this complex genetic architecture is important for predicting population variation and the evolutionary potential of the traits. Recently, theoretical work has identified specific patterns of gene interaction and gene effects across multiple traits that can have important consequences for evolutionary change. Yet, the classical tools for studying gene interactions are not well suited for detecting these patterns. There are also a lot of experimental data available that contain information about these patterns of genetic architecture, but for which there are no appropriate tools of statistical analysis.
The 'Statistical tools for studying genetic architectures (STSGA) project is aimed at developing statistical tools for extracting information about the genetic architecture that underlies biological traits from two types of data:
1) Artificial-selection time series (observations of trait values in subsequent generations in populations subject to controlled artificial selection), and
2) Quantitative trait loci (QTL) mapping data (i.e. molecular-marker based information of how differences at particular locations in the genome affect the traits).
Accordingly, there were two main research objectives:
1) Development of statistical methods to detect systematic patterns of epistatic gene interactions from marker data: We have achieved this objective in that we have developed and implemented a model for detecting genetic interactions (epistasis) in a freely-available software package. Specifically, this framework makes it possible to fit specific models that can detect systematic patterns of epistasis ('directional epistasis') that may have evolutionary implications. We have used this to analyse QTLs affecting the size of various bones and internal organs in mice, demonstrating regular patterns of gene interaction.
2) Development of likelihood-based methods to estimate genetic architecture from artificial selection response: We have achieved this objective in that we have developed a general statistical model framework that can be used to describe a variety of different genetic architectures, and then fit these to various types of quantitative genetic time-series data. This model is described in two publications, and we have several applications to real data sets at various stages of development.
The applications also include three analyses of selection on architecture in natural populations:
1) harvest-induced selection on pike,
2) natural selection on sticklebacks during invasion of freshwater,
3) selection on different female morphs in damselflies.
We have also developed a software package for this method that will be released during 2010.
The project has benefited from the collaboration with a number of outstanding researchers in Oslo and at other institutions including David Houle (Florida State University), Gunter Wagner (Yale University), Jim Cheverud (Washington University), Hans Skaug (University of Bergen), Jose Alvarez-Castro (Universidad Santiago de Compostela), Orjan Carlborg (Uppsala University), and many postdocs and faculty at the University of Oslo. We expect that the results achieved in this project will form the basis for many research publications over the coming years, and the project has been an excellent career step for the project researcher Arnaud Le Rouzic, who as a result has obtained a tenured position at the CNRS.
The 'Statistical tools for studying genetic architectures (STSGA) project is aimed at developing statistical tools for extracting information about the genetic architecture that underlies biological traits from two types of data:
1) Artificial-selection time series (observations of trait values in subsequent generations in populations subject to controlled artificial selection), and
2) Quantitative trait loci (QTL) mapping data (i.e. molecular-marker based information of how differences at particular locations in the genome affect the traits).
Accordingly, there were two main research objectives:
1) Development of statistical methods to detect systematic patterns of epistatic gene interactions from marker data: We have achieved this objective in that we have developed and implemented a model for detecting genetic interactions (epistasis) in a freely-available software package. Specifically, this framework makes it possible to fit specific models that can detect systematic patterns of epistasis ('directional epistasis') that may have evolutionary implications. We have used this to analyse QTLs affecting the size of various bones and internal organs in mice, demonstrating regular patterns of gene interaction.
2) Development of likelihood-based methods to estimate genetic architecture from artificial selection response: We have achieved this objective in that we have developed a general statistical model framework that can be used to describe a variety of different genetic architectures, and then fit these to various types of quantitative genetic time-series data. This model is described in two publications, and we have several applications to real data sets at various stages of development.
The applications also include three analyses of selection on architecture in natural populations:
1) harvest-induced selection on pike,
2) natural selection on sticklebacks during invasion of freshwater,
3) selection on different female morphs in damselflies.
We have also developed a software package for this method that will be released during 2010.
The project has benefited from the collaboration with a number of outstanding researchers in Oslo and at other institutions including David Houle (Florida State University), Gunter Wagner (Yale University), Jim Cheverud (Washington University), Hans Skaug (University of Bergen), Jose Alvarez-Castro (Universidad Santiago de Compostela), Orjan Carlborg (Uppsala University), and many postdocs and faculty at the University of Oslo. We expect that the results achieved in this project will form the basis for many research publications over the coming years, and the project has been an excellent career step for the project researcher Arnaud Le Rouzic, who as a result has obtained a tenured position at the CNRS.