Related Experiment Videos
Multivariate character process models for the analysis of two or more correlated function-valued traits.
Florence Jaffrézic1, Robin Thompson, Scott D Pletcher
1INRA Quantitative and Applied Genetics, 78352 Jouy-en-Josas Cedex, France. florence.jaffrezic@dga2.jouy.inra.fr
Genetics
|September 30, 2004
Summary
This study extends character process models for analyzing multiple correlated function-valued traits. Bivariate models effectively handle complex genetic covariance structures, outperforming other methods in fruit fly data analysis.
Area of Science:
- Quantitative Genetics
- Statistical Genetics
- Evolutionary Biology
Background:
- Longitudinal data and function-valued traits require specialized genetic analysis methods.
- Univariate character process models demonstrate strong performance for single traits.
- Existing methods may not adequately address the complexity of multiple correlated traits.
Purpose of the Study:
- To extend character process models for the simultaneous genetic analysis of two or more correlated function-valued traits.
- To investigate the analytical forms of stationary and nonstationary cross-covariance functions for these extended models.
- To compare the performance of bivariate character process models against alternative methods.
Main Methods:
- Development and analytical study of stationary and nonstationary cross-covariance functions for bivariate character process models.
- Simulation studies to compare bivariate character process models with random regression and structured antedependence models.
- Application of bivariate character process models to analyze genetic covariance in age-specific fecundity and mortality in Drosophila.
Main Results:
- Bivariate character process models with exponential correlation closely approximate first-order structured antedependence models.
- Methodological choice is highly dependent on the specific covariance structure of the data.
- Bivariate character process models successfully handle complex nonstationary and nonsymmetric cross-correlation structures.
Conclusions:
- The extended bivariate character process models provide a robust framework for analyzing correlated function-valued traits.
- These models are particularly suitable for complex genetic covariance structures, as demonstrated in the Drosophila example.
- The findings highlight the importance of selecting appropriate statistical methodologies based on data characteristics for accurate genetic analysis.